Current status of VTE risk assessment and prevention using clinical decision support system: a cross-sectional survey from China

doi:10.21203/rs.3.rs-5008620/v1

Download PDF

Research Article

Current status of VTE risk assessment and prevention using clinical decision support system: a cross-sectional survey from China

https://doi.org/10.21203/rs.3.rs-5008620/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background and Aim:

Venous thromboembolism (VTE) is an important cause of unexpected death in hospitalized patients. In recent years, Clinical Decision Support System (CDSS) has been increasingly adopted by hospitals worldwide. We conducted a survey with the aim of gaining a comprehensive understanding of the current state and future development direction of CDSS for VTE risk assessment and prevention(VTE-CDSS) in China.

Methods:

A network survey was conducted among hospitals in China. The investigation mainly included 39 questions, such as the implementation details of VTE-CDSS, the scale and the admission capacity of the hospitals. SPSS 20.0 software was used for statistical analysis.

Results:

A total of 587 hospitals responded to this survey, of which 194 (33.05%, 194/587) deployed VTE-CDSS, and less than a quarter (23.71%, 46/194) had Artificial intelligence(AI)-enabled VTE-CDSS. Among the 194 hospitals, the proportion of auxiliary decision support functions related to "risk assessment" was the highest (78.87%, 68.04%, 69.07%), followed by the auxiliary decision support functions related to "prophylaxis execution" (88.66%, 49.48%, 26.80%), and the proportion of auxiliary decision support functions related to "outcome event monitoring" was the lowest (46.39%, 22.68%). More than half of the respondents believed that the risk assessment rate and accuracy of the assessment had been significantly improved (56.19%, 109/194). However, only over one-third of respondents believed that the prevention rate and the standardization of prevention had been significantly improved (37.63%, 73/194). "The overall hospital information foundation is not perfect" was the primary hindrance factor in the implementation and application of VTE-CDSS (40.21%, 78/194). "System functions need to be further improved and more functional applications expanded" (78.35%, 152/194) is the most critical problem that VTE-CDSS needs to be further optimized and solved in the future. There were statistically significant differences between the two groups of hospitals that deployed VTE-CDSS with and without AI function (P<0.005) in the functional realization of various application details, the obstacles encountered in the implementation, and the problems to be further optimized and solved in the future. However, at the present stage in China, the efficiency and effectiveness of VTE-CDSS with AI function in risk assessment and implementation of prophylaxis measures are not significantly different from that of VTE-CDSS without AI function. (0.75<P<0.9).

Conclusions:

The information construction of VTE in China has developed rapidly. The implementation of VTE-CDSS achieved certain results, but there are still some obstacles and problems that need to be optimized in the future.

Survey

Venous thromboembolism (VTE)

Clinical Decision Support System (CDSS)

Artificial intelligence(AI)

Venous thromboembolism (VTE), defined as deep vein thrombosis(DVT), pulmonary thromboembolism (PTE), or both, has become an important issue of patient safety worldwide. VTE is associated with higher morbidity and mortality among hospitalized patients in low-, middle- and high-income countries. Furthermore, VTE is still considered as one of the main economic burdens on healthcare systems. Without appropriate prophylaxis, up to 25–60% of patients who undergo surgery develop a DVT that is detectable by venography [1]; for patients undergoing orthopedic surgery, it is known that if appropriate prophylaxis is used, symptomatic DVT and PTE can be reduced by 55–60% [2].

In recent years, international authoritative academic institutions have issued a series of clinical practice guidelines on the risk assessment and prevention of VTE. These guidelines are available to help risk-stratify patients into low, moderate, or high risk for VTE and provide the corresponding prophylaxis measure, including drug prophylaxis, mechanical prophylaxis, or both [2, 3, 4]. Despite the widespread availability of evidence-based clinical practice guidelines on VTE prophylaxis, however, clinicians do not uniformly use existing risk-stratification tools and, when used, clinicians often use the tools incorrectly, producing an underestimation of a patient’s risk for VTE [5, 6]. This underestimation leaves patients without adequate prophylaxis for VTE and significantly increases their risk for DVT or PTE. International multi-center research results show that the proportion of patients receiving appropriate VTE prevention in accordance with the guidelines is only 39.5–58.5% [7]. In China, only 9.0% of inpatients followed the guidelines for appropriate prevention, among which the rate of medical prevention was 6.0% and the rate of surgical prevention was 11.8% [8].

With the digital transformation of medicine and advances in health information technology, Clinical Decision Support Systems (CDSS) offer great promise to enhance efficiency and effectiveness of the prevention of VTE. A CDSS is rule or algorithm-based software that can analyze patient data and trigger timely, actionable, evidence based recommendations to healthcare teams to support clinical decisions [9]. Artificial intelligence (AI) technology has entered a phase of accelerated development in the 21st century, it has been used to solve medical problems in the fields of medical management, clinical decision support, patient monitoring, and health intervention. AI can be regarded as the emulation of human intelligence, which enables machines to learn from existing knowledge, adjust outputs, perform human-like tasks, and discover new knowledge. AI-based CDSS(AI + CDSS), which combines the knowledge reasoning technology of AI with the basic functional model of CDSS, has become a hot spot in the current medical field, and is considered to be an effective tool to improve the quality of medical care and enhance the work efficiency of clinicians [10].

In recent years, CDSS and AI + CDSS have been increasingly adopted by hospitals worldwide [11–13], including those in China [14–16], to support clinicians in VTE risk assessment and prevention. This implementation has yielded significant advancements in enhancing the accuracy and efficiency of risk assessment and prevention measures while concurrently reducing hospital-associated VTE incidents.

In October 2018, the official launch of the National VTE Prevention Program in China marked a significant emphasis on the establishment of a comprehensive VTE prevention management system and the utilization of information technology for VTE risk assessment and prevention in patients. To date, approximately 1,500 hospitals nationwide have actively participated in this program. However, the specific application of CDSS and AI + CDSS for VTE risk assessment and prevention in various hospitals in China remains unknown, including system functionalities, utilization effectiveness, implementation barriers, and areas for optimization and resolution. Therefore, we conducted a survey with the aim of gaining a comprehensive understanding of the current state and future development direction of CDSS and AI + CDSS for VTE risk assessment and prevention in China. Our goal is to identify targeted solutions through this research.

Survey Questionnaire Development and Pretesting

We consulted the existing literature on the application of CDSS for VTE risk assessment and prevention(VTE-CDSS), and integrate it with the current practical application situation in China. Collaborate with domestic experts in medical informatics (particularly those with research and practical experience in CDSS and AI), clinical specialists, healthcare management personnel, and information professionals to collectively develop the initial questionnaire.

Select ten representative hospitals from central, eastern, western, and northeastern regions of China that include large tertiary hospitals, medium-sized tertiary hospitals, as well as grassroots secondary hospitals. After conducting a pilot survey on a small scale, revise and enhance the survey questionnaire based on feedback received to finalize the questionnaire.

Contents of Survey Questionnaire

The contents of the questionnaire are mainly application status and future implementation suggestions for VTE-CDSS, which are composed of the following five parts (see supplemental table for details):

1. Covering letter;

2. Basic information of the surveyed hospitals: including the name, level, category, form of ownership, location, etc.;

3. Application and implementation of VTE-CDSS: including the development of VTE risk assessment and prevention work in hospitals, the specific application and implementation details of VTE-CDSS, utilization effectiveness, implementation barriers, and areas for optimization and resolution;

4. Scale of the surveyed hospitals: including the number of beds, the number of doctors and the number of nurses;

5. Admission capacity of the surveyed hospitals: annual outpatient volume, annual discharge number, annual operation volume;

6. Basic information of the respondents: including age, gender, title, education, work department, etc.

Survey Deployment

From September to December 2023, a network survey (electronic questionnaire) was adopted and forwarded by the National VTE Prevention Program Office of China to each provincial VTE prevention regional alliance, and the investigation purpose, survey objects and survey time period were explained in detail. Then, the provincial VTE prevention regional alliance forwarded the local hospitals, and each hospital assigned a responsible person who was familiar with the VTE-CDSS of the hospital to fill in the questionnaire.

Additionally, the National VTE Prevention Program Office of China, in collaboration with provincial VTE prevention regional alliances, sent at least 5 reminders through WeChat group announcements, emails, mobile phone text messages, and other communication channels to enhance the response rate of questionnaires.

Survey Sample Size

According to《Statistical Bulletin on the Development of Health Undertakings in China in 2022》[17]issued by the National Health Commission of the People's Republic of China, as of October 2023, there are a total of 14,668 hospitals in China, including 3,523 tertiary hospitals and 11,145 secondary hospitals. That is, the total N is 14,668.

Based on the sample size calculation formula: n=u²_α_/2π(1-π)/δ², with a 95% confidence interval and α set at 0.05, u=1.96; According to previous literature, the proportion of hospitals in China implementing CDSS is approximately 15%, with an overall rate of π=0.15. If the allowable error is set at 0.03, then δ=0.03. The calculated sample size n is 544.

The n/N ratio is below 0.05, which indicates that no correction to the sample size is necessary; therefore, the survey should include a minimum of 544 hospitals.

Survey Quality Control Measures

In the electronic questionnaire, we preseted a logical correlation between the questions and the specified mandatory responses. Prior to formal submission of the questionnaire, an automatic check for logical coherence and comprehensiveness was conducted. Before submission, respondents had the opportunity to review or modify their completed content.

We cleaned and organized the collected survey questionnaires, and removed 13 test questionnaires. Considering the IP address and hospital name, 11 hospitals were screened for duplicate questionnaires, and after cleaning, only the most authoritative one was retained as a valid questionnaire for each hospital. The cleaning rules are as follows:

1. Delete those with shorter filling times, such as deleting those that are completed within 30 seconds and retaining those that are completed within 500 seconds;

2. Completely identical, only keep one copy;

3. If there is a significant difference in the contents of the duplicate answer sheet, confirm to retain one copy after contacting the respondents.

Finally, a total of 611 questionnaires were collected in this survey. After excluding 13 test questionnaires and 11 repeated questionnaires, 587 effective questionnaires were obtained, with an effective rate of 96.07%.

Calculations and Statistical Analysis

Based on the survey results, we have summarized and statistically analyzed the demographic characteristics of the respondents (age, gender, professional title, education level, work department) and the basic information of the responding hospitals (economic region, grade, category and ownership); we calculated the proportion of responding hospitals performing VTE risk assessment and prevention, and the proportion of VTE-CDSS deployed in responding hospitals; we analyzed the functions and implementation details of VTE-CDSS in clinical practice; we also described the achievements, existing problems and obstacles of VTE-CDSS, and what needs to be further optimized.

We used a Chi-square test to compare whether there were differences in hospital background variables (economic region, hospital grade and category, hospital scale, admission capacity, etc.) between the two groups of VTE-CDSS with or without AI capabilities. Wilcoxon rank sum test was used to compare whether the two groups of VTE-CDSS with or without AI function had differences in implementation details and system effectiveness. Statistical significance was determined as p <0.05. Statistical calculation was performed by SPSS20.0 (IBM Corp. Released 2011, IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp.).

Geographical and Demographics Characteristics

The 587 valid questionnaires in this survey were sourced from 587 hospitals, of which 194 hospitals deployed VTE-CDSS (33.05%, 194/587). Therefore, the responses available for this survey were from 194 respondents in 194 hospitals across 27 provinces in China. The majority of respondents were aged from 31 to 40 years old (46.91%, 91/194), followed by 41 to 50 years old (31.96%, 62/194). The proportion of females (58.25%, 113/194) was slightly higher than that of males (41.75%, 81/194). The department they work in was mainly the medical affairs department (63.92%, 124/194), followed by the quality control department (15.46%, 30/194). The demographic characteristics of respondents included in this survey are detailed in Table 1.

Among the responding hospitals included in the analysis, a higher proportion (85.05%, 165/194) were tertiary hospitals compared to secondary hospitals, and general hospitals (95.36%, 185/194) outnumbered specialized hospitals. Furthermore, public hospitals (96.91%, 188/194) were more prevalent than private ones. Table 2 presents the geographical distribution and other fundamental characteristics of the participating hospitals.

Among the 194 responding hospitals included in this analysis, less than a quarter (23.71%, 46/194) had AI-enabled VTE-CDSS. The basic characteristics of hospitals with and without AI function VTE-CDSS were statistically analyzed, and it was found that there were no statistically significant differences between the two groups of hospitals in terms of economic regional distribution, hospital category and ownership. However, there were significant statistical differences in hospital scale (grade, number of beds, number of doctors, number of nurses) and admission capacity (annual outpatient volume, annual discharge number, annual operation volume). The p-values are shown in Table 2.

Therefore, we can conclude that the larger the scale and stronger the admission capacity of hospitals, the higher the proportion of deploying AI-enabled VTE-CDSS.

Table 1

Demographic characteristics of the respondents in this survey
Gender	N answers	Proportion
male	81	41.75%
female	113	58.25%
Age	N answers	Proportion
18 ~ 25	3	1.55%
26 ~ 30	22	11.34%
31 ~ 40	91	46.91%
41 ~ 50	62	31.96%
51 ~ 60	16	8.25%
≥ 60	0	0
Professional Title	N answers	Proportion
senior	27	13.92%
associate senior	56	28.87%
intermediate	70	36.08%
junior	41	21.13%
Educational Background	N answers	Proportion
bachelor	116	59.79%
master	73	37.63%
doctor	5	2.58%
Work Department	N answers	Proportion
neurosurgery	1	0.52%
cardiology	3	1.55%
orthopedics	3	1.55%
information technology department	3	1.55%
intensive care unit	1	0.52%
department of interventional medicine	4	2.06%
nursing department	8	4.12%
vascular surgery	8	4.12%
department of Pulmonary and Critical Care Medicine	9	4.64%
quality control department	30	15.46%
medical affairs department	124	63.92%

Table 2

Characteristics of responding hospitals in this survey
Characteristics	Total (194)	Hospitals with AI-enabled VTE-CDSS	Hospitals with non-AI-enabled VTE-CDSS	P-Value
Characteristics	Total (194)	(46, 23.71%)	(148, 76.29%)	P-Value
Economic region
Eastern	94(48.45%)	26(56.52%)	68(45.95%)	0.25 ~ 0.75
Midlands	44(22.68%)	7(15.22%)	37(25.00%)
Western	49(25.26%)	11(23.91%)	38(25.68%)
Northeast	7(3.61%)	2(4.35%)	5(3.38%)
Grade
Secondary hospital	29(14.95%)	2(4.35%)	27(18.24%)	0.01 ~ 0.025
Tertiary hospital	165(85.05%)	44(95.65%)	121(81.76%)	0.01 ~ 0.025
Category
General hospital	185(95.36%)	43(93.48%)	142(95.95%)	0.25 ~ 0.75
Specialized hospital	9(4.64%)	3(6.52%)	6(4.05%)	0.25 ~ 0.75
Ownership
Public	188(96.91%)	45(97.83%)	143(96.62%)	0.25 ~ 0.75
Private	6(3.09%)	1(2.17%)	5(3.38%)	0.25 ~ 0.75
Actual opening beds
< 500	45(23.20%)	6(13.04%)	39(26.35%)	0.005 ~ 0.01
500 ~ 999	53(27.32%)	7(15.22%)	46(31.08%)
1000 ~ 1499	28(14.43%)	12(26.09%)	16(10.81%)
1500 ~ 1999	56(28.87%)	16(34.78%)	40(27.03%)
≥ 2000	12(6.19%)	5(10.87%)	7(4.73%)
Number of doctors
< 100	16(8.25%)	1(2.17%)	15(10.14%)	< 0.005
100 ~ 199	32(16.49%)	5(10.87%)	27(18.24%)
200 ~ 499	76(39.18%)	10(21.74%)	66(44.59%)
500 ~ 999	25(12.89%)	8(17.39%)	17(11.49%)
≥ 1000	45(23.20%)	22(47.83%)	23(15.54%)
Number of nurses
< 500	30(15.46%)	5(10.87%)	25(16.89%)	0.005 ~ 0.01
500 ~ 999	75(38.66%)	9(19.57%)	66(44.59%)
1000 ~ 1499	36(18.56%)	14(30.43%)	22(14.86%)
1500 ~ 1999	30(15.46%)	11(23.91%)	19(12.84%)
≥ 2000	23(11.86%)	7(15.22%)	16(10.81%)
Outpatient quantity per year
< 100,000	48(24.74%)	7(15.22%)	41(27.70%)	0.025 ~ 0.05
100,000 to 500,000	51(26.29%)	8(17.39%)	43(29.05%)
500,000 to 1 million	24(12.37%)	10(21.74%)	14(9.46%)
1 million to 2 million	57(29.38%)	17(36.96%)	40(27.03%)
≥ 2 million	14(7.22%)	4(8.70%)	10(6.76%)
Discharged patients’ quantity per year
< 20,000	21(10.82%)	3(6.52%)	18(12.16%)	< 0.005
20,000 to 50,000	32(16.49%)	6(13.04%)	26(17.57%)
50,000 to 100,000	27(13.92%)	8(17.39%)	19(12.84%)
100,000 to 200,000	72(37.11%)	9(19.57%)	63(42.57%)
≥ 200,000	42(21.65%)	20(43.48%)	22(14.86%)
Surgery quantity per year
< 10,000	50(25.77%)	8(17.39%)	42(28.38%)	0.005 ~ 0.01
10,000 to 20,000	53(27.32%)	6(13.04%)	47(31.76%)
20,000 to 50,000	57(29.38%)	18(39.13%)	39(26.35%)
50,000 to 100,000	24(12.37%)	10(21.74%)	14(9.46%)
≥ 100,000	10(5.15%)	4(8.70%)	6(4.05%)

Legend: All answers are reported as N (proportion in %). VTE Venous thromboembolism, CDSS Clinical Decision Support System, AI Artificial intelligence.

Implementation details of VTE-CDSS

In order to understand the specific functions of VTE-CDSS in China, we conducted statistical analysis on the application details at three aspects: CDSS assisted risk assessment, CDSS assisted prophylaxis implementation, and CDSS assisted endpoint event monitoring, as shown in Table 3.

Among the 194 responding hospitals included in this analysis, the proportion of auxiliary decision support functions related to "risk assessment" was the highest (78.87%, 68.04%, 69.07%), followed by the auxiliary decision support functions related to "prophylaxis execution" (88.66%, 49.48%, 26.80%), and the proportion of auxiliary decision support functions related to "outcome event monitoring" was the lowest (46.39%, 22.68%). Among them, "the system can recommend prophylaxis measures based on VTE risk and bleeding risk" (88.66%, 172/194) achieved the highest proportion of VTE-CDSS in 194 hospitals, while "the system can automatically identify cases without adequate prophylaxis for VTE and give early warning " (26.80%, 52/194) and "the system can automatically screen bleeding events related to VTE prevention or treatment " (22.68%, 44/194) achieved a relatively low proportion in the VTE-CDSS of 194 hospitals.

For hospitals that have deployed VTE-CDSS with and without AI function, statistical analysis was conducted on the implementation details of the above applications. Significant statistical differences were found between the two groups of hospitals, with p-values shown in Table 3. Therefore, we can conclude that VTE-CDSS with AI function is superior to VTE-CDSS without AI function in terms of functional realization of various application details.

Table 3

Implementation details of VTE-CDSS
Implementation details of VTE-CDSS	Total (194)	Hospitals with AI-enabled VTE-CDSS	Hospitals with non-AI-enabled VTE-CDSS	P-Value
Implementation details of VTE-CDSS	Total (194)	(46, 23.71%)	(148, 76.29%)	P-Value
Implementation details of CDSS assisted risk assessment				< 0.005
The system can automatically assess the risk of VTE and give early warning to intermediate and high risk cases of VTE	153(78.87%)	45(97.83%)	108(72.97%)
The system can automatically assess the bleeding risk and alert high-risk cases of bleeding	132(68.04%)	43(93.48%)	89(60.14%)
The system can automatically identify key dynamic time points and provide warning prompts for unassessed cases	134(69.07%)	43(93.48%)	91(61.49%)
Implementation details of CDSS assisted prophylaxis measures
The system can recommend prophylaxis measures based on VTE risk and bleeding risk	172(88.66%)	43(93.48%)	129(87.16%)
The system can automatically identify cases where prophylaxis measures have not been taken and provide warning prompts	96(49.48%)	28(60.87%)	68(45.95%)
The system can automatically identify cases without adequate prophylaxis for VTE and give early warning	52(26.80%)	23(50.00%)	29(19.59%)
Implementation details of CDSS assisted outcome event monitoring
The system can automatically screen hospital-associated VTE events	90(46.39%)	31(67.39%)	59(39.86%)
The system can automatically screen bleeding events related to VTE prevention or treatment	44(22.68%)	18(39.13%)	26(17.57%)

Legend: All answers are reported as N (proportion in %). VTE Venous thromboembolism, CDSS Clinical Decision Support System, AI Artificial intelligence.

Implementation status of VTE-CDSS assisted in monitoring quality indicators

This survey conducted a statistical analysis of the VTE-CDSS assisted monitoring of three types of core quality indicators (including risk assessment indicators, prevention indicators, and outcome indicators), as shown in Table 4.

Among the 194 responding hospitals included in this analysis, the proportion of VTE-CDSS capable of auxiliary monitoring of risk assessment indicators was the highest (82.47%-96.39%), followed by prevention indicators (87.63%-91.24%), and the proportion of auxiliary monitoring of outcome indicators was the lowest (53.61%-86.08%). Among the three types of core quality indicators, the degree of auxiliary monitoring of "intrinsic quality indicators" is relatively low, such as the accuracy of risk assessment (34.02% of hospitals, 66/194), the standardization of prophylaxis (37.63% of hospitals, 73/194), and the standardized treatment rate of hospital-associated VTE (22.68% of hospitals, 73/194).

There were significant statistical differences in the implementation of auxiliary monitoring functions for three core quality indicators between two groups of hospitals deployed with and without AI enabled VTE-CDSS, with p-values shown in Table 4. Therefore, we can conclude that VTE-CDSS with AI function is superior to VTE-CDSS without AI function in achieving auxiliary monitoring functions of each core quality index.

Table 4

Implementation status of VTE-CDSS assisted in monitoring quality indicators
Implementation status of VTE-CDSS assisted in monitoring quality indicators	Total (194)	Hospitals with AI-enabled VTE-CDSS	Hospitals with non-AI-enabled VTE-CDSS	P-Value
	Total (194)	(46, 23.71%)	(148, 76.29%)	P-Value
Risk assessment indicators which the VTE-CDSS can assisted in monitoring
VTE risk assessment rate	187(96.39%)	46(100.00%)	141(95.27%)	< 0.005
Proportion of moderate to high risk of VTE	183(94.33%)	46(100.00%)	137(92.57%)
Bleeding risk assessment rate	184(94.85%)	46(100.00%)	138(93.24%)
Proportion of high risk of bleeding	160(82.47%)	45(97.83%)	115(77.70%)
Accuracy of risk assessment*	66(34.02%)	21(45.65%)	45(30.41%)
Prevention indicators which the VTE-CDSS can assisted in monitoring
Drug prophylaxis rate	177(91.24%)	46(100.00%)	131(88.51%)	< 0.005
Mechanical prophylaxis rate	176(90.72%)	45(97.83%)	131(88.51%)
Combined prophylaxis rate	170(87.63%)	45(97.83%)	125(84.46%)
Standardization of prophylaxis *	73(37.63%)	23(50.00%)	50(33.78%)
Outcome indicators which the VTE-CDSS can assisted in monitoring
Incidence of hospital-associated VTE	167(86.08%)	43(93.48%)	124(83.78%)	< 0.005
Standardized treatment rate of hospital-associated VTE *	44(22.68%)	31(67.39%)	13(8.78%)
Incidence of bleeding events	104(53.61%)	33(71.74%)	71(47.97%)
Mortality of hospital-associated VTE	164(84.54%)	45(97.83%)	119(80.41%)

Legend: All answers are reported as N (proportion in %). VTE Venous thromboembolism, CDSS Clinical Decision Support System, AI Artificial intelligence.

*: intrinsic quality indicators

The efficiency and effectiveness of using VTE-CDSS

In this survey, participants were consulted for their opinions and evaluations on the effectiveness of VTE-CDSS in clinical practice, as shown in Table 5.

Among the 194 respondents included in this analysis, nearly 3/4 of them believed that VTE-CDSS had "perfect system functions, and high enthusiasm for use by medical personnel" (73.71%, 143/194). Among responders with AI enabled VTE-CDSS deployed in their hospital, the highest proportion (82.61%, 38/46) believed that "perfect system functions, and high enthusiasm for use by medical personnel". However, hospitals that deploy AI-enabled VTE-CDSS had a higher demand for vendor intervention support compared to hospitals with non-AI-enabled VTE-CDSS (21.74% vs. 18.24%).

More than half of the respondents believed that the risk assessment rate and accuracy of the assessment had been significantly improved (56.19%, 109/194). However, only over one-third of respondents believed that the prevention rate and the standardization of prevention had been significantly improved (37.63%, 73/194). There was no significant statistical difference in the efficiency and effectiveness of risk assessment and implementation of prophylaxis measures between two groups of hospitals deployed with and without AI enabled VTE-CDSS, and the P-values are shown in Table 5.

Therefore, we can conclude that at the present stage in China, the efficiency and effectiveness of VTE-CDSS with AI function in risk assessment and implementation of prophylaxis measures are not significantly different from that of VTE-CDSS without AI function. Although VTE-CDSS with AI function is more complete in system functionality and easier to accept and use by medical personnel than those without AI function, it may require more intervention and support from manufacturers in system debugging and integration.

Table 5

The efficiency and effectiveness of using VTE-CDSS
The efficiency and effectiveness of using VTE-CDSS	Total (194)	Hospitals with AI-enabled VTE-CDSS	Hospitals with non-AI-enabled VTE-CDSS	P-Value
The efficiency and effectiveness of using VTE-CDSS	Total (194)	(46, 23.71%)	(148, 76.29%)	P-Value
Evaluation of system function
Perfect system functions, and high enthusiasm for use by medical personnel	143(73.71%)	38(82.61%)	105(70.95%)	< 0.005
Perfect system functions, but the number of users is small	24(12.37%)	5(10.87%)	19(12.84%)
Incomplete system functions, and the number of users is not large	13(6.70%)	2(4.35%)	11(7.43%)
System administrator intervention is often required	48(24.74%)	8(17.39%)	40(27.03%)
Vender intervention is often required	37(19.07%)	10(21.74%)	27(18.24%)
Efficiency and effectiveness——Risk assessment
The risk assessment rate and accuracy of the assessment have been significantly improved (the improvement rate is more than 80%)	109(56.19%)	27(58.70%)	82(55.41%)	0.75 ~ 0.9
The risk assessment rate and accuracy of the assessment have been significantly improved (the improvement rate is more than 30%)	59(30.41%)	12(26.09%)	47(31.76%)
The assessment rate has improved, but the accuracy of the assessment has not	22(11.34%)	6(13.04%)	16(10.81%)
Effect uncertainty	4(2.06%)	1(2.17%)	3(2.03%)
Efficiency and effectiveness——Prophylaxis
The prevention rate and the standardization of prevention have been significantly improved (the improvement rate is more than 80%)	73(37.63%)	20(43.48%)	53(35.81%)	0.75 ~ 0.9
The prevention rate and the standardization of prevention have been improved to some extent (the improvement rate is more than 30%)	87(44.85%)	19(41.30%)	68(45.95%)
The prevention rate has improved, but the standardization of prevention has not	28(14.43%)	6(13.04%)	22(14.86%)
Effect uncertainty	6(3.09%)	1(2.17%)	5(3.38%)

Legend: All answers are reported as N (proportion in %). VTE Venous thromboembolism, CDSS Clinical Decision Support System, AI Artificial intelligence.

Barriers in using VTE-CDSS

This survey conducted a statistical analysis on the obstacles existing in the implementation of VTE-CDSS, as shown in Table 6. Among them, "the overall hospital information foundation is not perfect" accounted for the highest proportion (40.21%, 78/194), followed by "VTE-CDSS lacks effective suggestions of prophylaxis measures to guide clinical practice" (34.02%, 66/194), "VTE-CDSS has poor compatibility" (30.93%, 60/194), "the informationization process between doctors and nurses is not smooth" (24.74%, 48/194), etc.

It is worth noting that the following hindrance factors, such as "changing the work habits of doctors and nurses, increasing workload and affecting work efficiency", "excessive warning and prompt information interferes with normal clinical work", "doctors and nurses overly rely on VTE-CDSS, losing their independent thinking ability and clinical autonomy, inducing doctor-patient conflicts and affecting doctor-patient relationships", "warning and prompt information is useless or even incorrect", although the proportion is not high, it indicates that the system function of VTE-CDSS urgently needs to be closely adapted to clinical practice and further optimized.

There was a significant statistical difference between the two groups of hospitals that deployed VTE-CDSS with and without AI function in the obstacles encountered in the implementation, and the P-values were shown in Table 6. The proportion of most hindrance factors that occur in hospitals with AI-enabled function VTE-CDSS is higher than that in hospitals with non-AI-enabled VTE-CDSS. However, the proportion of these three obstacles, such as "the content of knowledge base is not updated in a timely manner and lacks authority", "doctors and nurses overly rely on VTE-CDSS, losing their independent thinking ability and clinical autonomy, inducing doctor-patient conflicts and affecting doctor-patient relationships", "warning and prompt information is useless or even incorrect", appearing in hospitals with AI-enabled VTE-CDSS is higher than that in hospitals with non-AI-enabled VTE-CDSS.

Table 6

Barriers in using VTE-CDSS
Barriers in using VTE-CDSS	Total	Hospitals with AI-enabled VTE-CDSS	Hospitals with non-AI-enabled VTE-CDSS	P-Value
Barriers in using VTE-CDSS	(194)	(46, 23.71%)	(148, 76.29%)	P-Value
The overall hospital information foundation is not perfect	78(40.21%)	14(30.43%)	64(43.24%)	< 0.005
VTE-CDSS lacks effective suggestions of prophylaxis measures to guide clinical practice	66(34.02%)	14(30.43%)	52(35.14%)
VTE-CDSS has poor compatibility	60(30.93%)	11(23.91%)	49(33.11%)
The informationization process between doctors and nurses is not smooth	48(24.74%)	11(23.91%)	37(25.00%)
The content of knowledge base is not updated in a timely manner and lacks authority	44(22.68%)	11(23.91%)	33(22.30%)
VTE-CDSS costs too much	42(21.65%)	9(19.57%)	33(22.30%)
Changing the work habits of doctors and nurses, increasing workload and affecting work efficiency	34(17.53%)	6(13.04%)	28(18.92%)
Excessive warning and prompt information interferes with normal clinical work	31(15.98%)	7(15.22%)	24(16.22%)
Doctors and nurses overly rely on VTE-CDSS, losing their independent thinking ability and clinical autonomy, inducing doctor-patient conflicts and affecting doctor-patient relationships	17(8.76%)	5(10.87%)	12(8.11%)
Warning and prompt information is useless or even incorrect	8(4.12%)	2(4.35%)	6(4.05%)

Legend: All answers are reported as N (proportion in %). VTE Venous thromboembolism, CDSS Clinical Decision Support System, AI Artificial intelligence.

Problems to be further optimized and solved in the future in the implementation of VTE-CDSS

This survey conducted a statistical analysis on the problems that VTE-CDSS needs to be further optimized and solved in the future, as shown in Table 7.

Among the 194 respondents included in this analysis, over 70% believed that "system functions need further improvement and more functional applications need to be expanded" (78.35%, 152/194), and "Clinical pathways for VTE risk assessment and prevention need to be smoothly integrated with CDSS" (73.20%, 142/194). Compared to hospitals with AI-enabled VTE-CDSS, hospitals with non-AI-enabled VTE-CDSS had more urgent needs for the above two items (81.08% vs 69.57%, 74.32% vs 69.57%).

More than 50% of respondents believed that "data quality and availability need to be further improved", "VTE-CDSS construction standards and application management policies need to be developed", "AI algorithms, such as interpretability, transparency, algorithm adaptability, etc., need to be further improved", "unified VTE dataset standards need to be established and promoted nationwide", "more user-friendly human-computer interaction interfaces" are all issues that need to be further optimized and solved in the future.

As for the problems that need to be further optimized and solved in the future, there was a significant statistical difference between two groups of hospitals that have deployed VTE-CDSS with and without AI function, and the P-values were shown in Table 7.

Table 7

Problems to be further optimized and solved in the future
Problems to be further optimized and solved in the future	Total	Hospitals with AI-enabled VTE-CDSS	Hospitals with non-AI-enabled VTE-CDSS	P-Value
Problems to be further optimized and solved in the future	(194)	(46, 23.71%)	(148, 76.29%)	P-Value
System functions need further improvement and more functional applications need to be expanded	152(78.35%)	32(69.57%)	120(81.08%)	< 0.005
Clinical pathways for VTE risk assessment and prevention need to be smoothly integrated with CDSS	142(73.20%)	32(69.57%)	110(74.32%)
Data quality and availability need to be further improved	114(58.76%)	29(63.04%)	84(56.76%)
VTE-CDSS construction standards and application management policies need to be developed	108(55.67%)	28(60.87%)	80(54.05%)
AI algorithms, such as interpretability, transparency, algorithm adaptability, etc., need to be further improved	106(54.64%)	30(65.22%)	76(51.35%)
Unified VTE dataset standards need to be established and promoted nationwide	105(54.12%)	29(63.04%)	76(51.35%)
More user-friendly human-computer interaction interfaces	99(51.03%)	29(63.04%)	70(47.30%)
Increase objective data on user usage	81(41.75%)	21(45.65%)	60(40.54%)
Reduce system implementation workload	72(37.11%)	18(39.13%)	54(36.49%)

Legend: All answers are reported as N (proportion in %). VTE Venous thromboembolism, CDSS Clinical Decision Support System, AI Artificial intelligence.

With the promotion of the National VTE Prevention Program in China, the information construction of VTE has developed rapidly in recent years. Especially since 2018, the number of hospitals deploying VTE-CDSS has increased significantly, and the system functions have been gradually shifted from basic risk assessment computerization to more in-depth and diverse functions such as process quality control and monitoring, and AI-assisted clinical decision-making. The implementation of VTE-CDSS in China has achieved certain results, but there are still some obstacles and problems that need to be optimized in the future.

Improving system functionality to adapt to clinical pathways for VTE risk assessment and prevention

The biggest challenge faced by VTE-CDSS in the implementation process of this survey is to further improve system functionality to align with the clinical pathway of VTE risk assessment and prevention. This is consistent with the issue raised by other investigations in China [18], which suggests that there may be a mismatch between VTE-CDSS system functionality and clinical business needs, making it difficult to translate user needs into system functionality design and closely align with clinical practice. This is also one of the main obstacles faced by VTE-CDSS in the process of system implementation.

Therefore, it is recommended that throughout the entire lifecycle of VTE-CDSS (design, development, implementation, operation and optimization), medical professionals and R&D personnel jointly participate in the creation of CDSS system solutions[19], taking more consideration of the application scope and service scenarios of the system, combining with hospital reality and clinical pathways of VTE risk assessment and prevention, closely communicating with medical professionals, and integrating the true needs of users and the pain points and difficulties of clinical business into the development of the system, so as to improve the clinical fit of the system. On the basis of breaking through the technical barriers, the standardized VTE-CDSS computerized clinical path is promoted from top to bottom, truly improving medical quality and clinical efficiency.

Deeply understand and implement the "Five Correct" principles of VTE-CDSS

Although the application of VTE-CDSS in hospitals in China is not yet widespread (33.05%, 194/587), medical personnel have recognized and appreciated its potential benefits and possible challenges in the future.

In this survey, the functional implementation of VTE-CDSS is mainly focused on assisting risk assessment and implementing prophylaxis measures. However, at the same time, we also concerned that the lack of effective prophylaxis measures to guide clinical practice is the second major obstacle to VTE-CDSS. A small number of respondents also mentioned that VTE-CDSS had "too much warning information, interfering with normal clinical work", and "warning information is useless or even incorrect".

The implementation of CDSS should follow the "Five Corrects" principle [20]: providing the right information to the right people through the right channels, at the right time and in the right intervention mode in the diagnosis and treatment process. The clinical expertise involved in VTE-CDSS is relatively intensive, which requires the R&D team to deeply understand what is the right time, who is the right person, and what prophylaxis measure recommendations are correct and effective information in combination with the clinical business process of VTE risk assessment and prevention. More research and practice is still needed to seamlessly integrate CDSS into clinical business processes for VTE risk assessment and prevention, where, when and in what intervention mode, which can effectively improve the execution of prophylaxis measures, reduce unnecessary information interference, and minimize the burden of system response.

While dynamically integrating a comprehensive and evidence-based medical knowledge base, VTE-CDSS needs to conduct medical logical training and conditional weight analysis training tailoring to the clinical characteristics of VTE risk assessment and prevention. This enables CDSS to truly understand medical records and VTE prophylaxis guidelines, accurately grasp the "Five Correct" implementation principles of CDSS, and empower medical personnel with a higher level of risk assessment and prevention capabilities.

Promote the application of VTE standard datasets to improve data quality and availability

"Data quality and availability", which has attracted more and more global attention, is also a key issue that all hospitals in this survey believe that VTE-CDSS needs to be further optimized and solved in the future.

The risk assessment and prevention of VTE involve a wide range of patients, covering almost all inpatient departments in the hospital. However, VTE risk assessment and prevention need to comprehensively consider patient factors, underlying diseases, concomitant medications, blood coagulation function and invasive operations, etc. [21]. These key medical information are scattered throughout various information systems within the hospital. Currently, the information systems within a hospital in China are often developed by multiple different software vendors. Systems developed by different suppliers have large structural differences in their technical architecture and data, and hospitals generate massive amounts of operation data and disease data every day, resulting in the integrated application of VTE-CDSS, which requires a significant investment of manpower and financial resources for data cleaning, governance, and standardization. On the other hand, in order to achieve a high degree of accuracy in assisted decision-making, AI-enabled VTE-CDSS often needs to conduct cross-center and cross-regional data collection to obtain sufficient evidence-based medicine sample data to support machine learning, which also puts higher requirements on the standardization, structuring and unification of data models [22].

To change this situation, it is urgent to establish a VTE standard dataset based on VTE's own disease characteristics and combined with existing industry standards, so as to lay a technical foundation for future VTE clinical normative management, real-world research, and nationwide quality control. Based on this, the National VTE Prevention Program Office of China organized experts in relevant fields to integrate existing terminology norms, clinical practice guidelines and expert consensus from the perspective of data management, conducted a comprehensive and systematic collection of VTE-related data elements, and compiled and published the "Venous Thromboembolism Standard Dataset" in January 2023.

We look forward to the nationwide promotion and application of the " Venous Thromboembolism Standard Dataset " to effectively promote the quality control of VTE and the integration and utilization of clinical data resources, standardize the management of VTE data in China, provide data support for the application of AI-based VTE-CDSS, and promote the standardized and homogeneous development of VTE prophylaxis and treatment system construction in China.

Improve the transparency and interpretability of VTE-CDSS related AI algorithms

In recent years, the application of AI-based CDSS in the VTE field has become increasingly active. In this survey, although the number of hospitals deploying AI-enabled VTE-CDSS was relatively small (7.84%, 46/587), more than half of the surveyed hospitals (54.64%, 106/194) are highly concerned about the interpretability and transparency of AI algorithms. This will be another important challenge for AI-based VTE-CDSS in the future.

In healthcare, the interpretability and transparency of AI is particularly important. Both patients and medical staff are not solely satisfied with the medical outcomes of AI output, but are also eager to understand the reasons behind making clinical decisions. Healthcare workers' understanding of how AI algorithms work, AI-prompted risk assessment and prophylaxis decision information affect their trust in the system. However, many machine learning methods ("black boxes") lack transparency, which may undermine the trust of healthcare professionals in AI output results [23,24].

To solve these problems, the R&D team needs to consider how to open the "black box" of AI algorithms, such as developing interpretable AI algorithms or interactive human-computer dialogue software, so that medical personnel can understand the logic behind a clinical decision suggestion and judge its accuracy [19, 25]. In addition, AI-based VTE-CDSS should proactively provide medical personnel with the details of risk assessment and prevention decisions, as well as relevant evidence-based medical evidence, such as source information support for patients' VTE risk factors and clinical evidence behind VTE prophylaxis measures recommendations, so that medical personnel can quickly understand the basis for clinical decision-making in a short time. This may be more favored by healthcare professionals than other software products with equal accuracy.

Strengths and Potential Limitations

In summary, this survey on the application of CDSS for VTE risk assessment and prevention in China has two advantages. Firstly, to our knowledge, this is the first study to investigate the system functionality, effectiveness, and barriers to implementation of VTE-CDSS, and to evaluate the differences between deploying VTE-CDSS with and without AI capabilities. Our investigation has received widespread response and acceptance in China. Secondly, 79.38% of the respondents in our survey were administrative personnel from the medical affairs department or quality control department, who were responsible for the overall deployment and specific implementation of VTE-CDSS in their hospitals, so they had a more comprehensive understanding of the application of VTE-CDSS in hospitals. This will help us to accurately grasp the current status and potential challenges of VTE-CDSS application and implementation in China.

On the other hand, our online survey also has some potential limitations. Firstly, as we did not have a tracking system to distinguish responders from non-responders, we were unable to compare detailed information about those who were not included in the survey and their hospitals, so we did not analyze the impact of non-respondent bias. This non-response bias may have some adverse effects on the representativeness of survey results. Because the respondents were likely to have a special interest in VTE-CDSS, most respondents hold a positive attitude toward VTE-CDSS. Secondly, quantitative studies may not be sufficient to fully understand the current application status and implementation barriers of VTE-CDSS. In fact, focusing solely on technology while ignoring human factors may lead to further explanatory oversights. Considering the social attribute of this topic, a combination of qualitative and quantitative methods in research may provide additional understanding of this complex phenomenon, which we will consider in our future work.

The application of CDSS and AI + CDSS for VTE risk assessment and prevention, although not yet widespread in hospitals in China, has made certain progress and breakthroughs in the implementation details of auxiliary risk assessment and prophylaxis measures, auxiliary quality control and monitoring, etc. Its effectiveness in improving the risk assessment rate, the accuracy of risk assessment, the prevention rate and the standardization of prevention has also been recognized by most users. The popularization and promotion of VTE-CDSS in Chinese hospitals still need to overcome some external and internal obstacles, and further optimize and solve the key issues that users are concerned about. Although facing challenges in the future, there is still great potential and development space.

AI	Artificial intelligence
AI+CDSS	AI-based Clinical Decision Support Systems; AI-based CDSS
CDSS	Clinical Decision Support Systems
DVT	Deep vein thrombosis
PTE	Pulmonary thromboembolism
VTE	Venous thromboembolism
VTE-CDSS	CDSS for VTE risk assessment and prevention
R&D	Research and development

Acknowledgements

We thank all the professionals who offered precious suggestions to this study. We thank you survey participants for their valuable input and help in improving this study. We thank all the healthcare workers who contributed to the data collection and management.

Authors’ contributions

CJ and ZZ contributed to the research idea and study design. CS, YT, LJ, JX, YX, WW, LZ and WX developed and designed the initial questionnaire. JW, YY, ZL, YJ, JS, QY, GS, YG, NZ, ZC, LZ, and ZC participated in the pilot survey and provided critical questionnaire review. QG, MS and BL were involved in data acquisition. LX, KZ, ZC, RL and XZ performed the statistical analysis and data interpretation. XZ and DW directed the statistical methods. The first draft of the manuscript was written by LX and all authors commented on previous versions of the of the paper. CJ and ZZ had full access to all data in the study and verified the data and are responsible for the integrity and accuracy of the data and the decision to submit the manuscript. All authors revised the report and approved the final version before submission.

Funding

This study is supported by National High Level Hospital Clinical Research Funding (2022‑NHLHCRF‑LX‑01‑0108); the CAMS Innovation Fund for Medical Sciences (CIFMS) (2023-I2M-A-014); National Key Research and Development Program of China (2023YFC2507200).

Availability of data and materials

According to the Personal Information Protection Law of China, individual participant data in our study will not be made available publicly. The data from the National VTE Prevention Program in China, will be made available upon publication to members of the scientific and medical community for non-commercial use only, upon email request to [email protected].

Ethics approval and consent to participate

This study was approved by the Clinical Research Ethical Committee of the science and technology center of China-Japan Friendship Hospital (approval number:2021-162-K120). All participants gave informed consent prior to taking the survey. Participation was voluntary and the data collected was nonidentifiable.

Clinical trial number：not applicable

Competing interests

The authors have no conflict of interest or financial relationships to disclose. No form of payment was given to anyone to produce the manuscript.

Geerts WH, Bergqvist D, Pineo GF, et al; American College of Chest Physicians. Prevention of venous thromboembolism: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest. 2008;133(6) (suppl):381S-453S.
Falck-Ytter Y, Francis CW, Johanson NA, et al; American College of Chest Physicians. Prevention of VTE in orthopedic surgery patients: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(2) (suppl): e278S-e325S.
Gould MK, Garcia DA, Wren SM, et al; American College of Chest Physicians. Prevention of VTE in nonorthopedic surgical patients: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(2) (suppl): e227S-e277S.
Caprini JA, Tapson VF, Hyers TM, et al; NABOR Steering Committee. Treatment of venous thromboembolism: adherence to guidelines and impact of physician knowledge, attitudes, and beliefs. J Vasc Surg. 2005;42(4):726-733.
Beck MJ, Haidet P, Todoric K, Lehman E, Sciamanna C. Reliability of a point-based VTE risk assessment tool in the hands of medical residents. J Hosp Med. 2011;6(4):195-201.
Pannucci CJ, Obi A, Alvarez R, et al. Inadequate venous thromboembolism risk stratification predicts venous thromboembolic events in surgical intensive care unit patients. J Am Coll Surg. 2014; 218(5):898-904.
Cohen AT, Tapson VF, Bergmann JF, et al; ENDORSE Investigators. Venous thromboembolism risk and prophylaxis in the acute hospital care setting (ENDORSE study): a multinational cross-sectional study. Lancet. 2008;371(9610):387-394.
Zhai Z, Kan Q, Li W et al. VTE Risk Profiles and Prophylaxis in Medical and Surgical Inpatients: The Identification of Chinese Hospitalized Patients' Risk Profile for Venous Thromboembolism (DissolVE-2)-A Cross-sectional StudyJ. . Chest,2019,155(1): 114-122.DOI: 10.1016/j.chest.2018.09.020.
Moja L, Kwag KH, Lytras T, et al. Effectiveness of computerized decision support systems linked to electronic health records: a systematic review and meta-analysis. Am J Public Health 2014;104: e12-22. DOI:10.2105/AJPH.2014.302164.
Nakamura T, Sasano T. Artificial intelligence and cardiology: current status and perspectiveJ. . J Cardiol 2022; 79:326–33. https://doi.org/10.1016/j.jjcc.2021.11.017.
Titi MA, Alotair HA, Fayed A, et al. Effects of Computerised Clinical Decision Support on Adherence to VTE Prophylaxis Clinical Practice Guidelines among Hospitalised Patients. Int J Qual Health Care. 2021;33(1): mzab034. DOI:10.1093/intqhc/mzab034.
Borab ZM, Lanni MA, Tecce MG, Pannucci CJ, Fischer JP. Use of Computerized Clinical Decision Support Systems to Prevent Venous Thromboembolism in Surgical Patients: A Systematic Review and Meta-analysis. JAMA Surg. 2017;152(7):638-645. DOI:10.1001/jamasurg.2017.0131.
Haut ER, Owodunni OP, Wang J, et al. Alert-Triggered Patient Education Versus Nurse Feedback for Nonadministered Venous Thromboembolism Prophylaxis Doses: A Cluster-Randomized Controlled Trial. J Am Heart Assoc. 2022;11(18): e027119. DOI:10.1161/JAHA.122.027119.
Huang X, Zhou S, Ma X, et al. Effectiveness of an artificial intelligence clinical assistant decision support system to improve the incidence of hospital-associated venous thromboembolism: a prospective, randomised controlled study. BMJ Open Qual. 2023;12(4): e002267. DOI:10.1136/bmjoq-2023-002267.
Gao Q, Zhen K, Xia L, et al. Assessment of the Effect on Thromboprophylaxis with Multifaceted Quality Improvement Intervention based on Clinical Decision Support System in Hospitalized Patients: A Pilot Study. J Clin Med. 2022;11(17):4997. Published 2022 Aug 25. DOI:10.3390/jcm11174997.
Jin ZG, Zhang H, Tai MH, Yang Y, Yao Y, Guo YT. Natural Language Processing in a Clinical Decision Support System for the Identification of Venous Thromboembolism: Algorithm Development and Validation. J Med Internet Res. 2023;25: e43153. Published 2023 Apr 24. DOI:10.2196/43153.
The National Health Commission of the People's Republic of China. Statistical Bulletin on the Development of Health Undertakings in China in 2022. EB/OL. . (2023-10-12) 2023-10-12. . http://www.nhc.gov.cn/guihuaxxs/s3586s/202310/5d9a6423f2b74587ac9ca41ab0a75f66.shtml.
Ji M, Chen X, Georgi Z Genchev, et al. Status of AI-Enabled Clinical Decision Support Systems Implementations in China. Methods Inf Med. 2021;60(5-06):123-132. DOI:10.1055/s-0041-1736461.
Liu S, McCoy AB, Peterson JF, et al. Leveraging explainable artificial intelligence to optimize clinical decision support. J Am Med Inform Assoc. Published online February 22, 2024. DOI:10.1093/jamia/ocae019.
Agency for Healthcare Research and Quality. A Guide for Effective Quality Improvement: Preventing Hospital-Acquired Venous Thromboembolism. 2nd EdEB/OL. . (2016-08-01) 2022-06-01. . https://www.ahrq.gov/patient-safety/resources/vtguide/index.html.
Henke PK, Kahn SR, Pannucci CJ, et al. Call to Action to Prevent Venous Thromboembolism in Hospitalized Patients: A Policy Statement From the American Heart Association. Circulation. 2020;141(24): e914-e931. DOI:10.1161/CIR.0000000000000769
Lu Y, Melnick ER, Krumholz HM. Clinical decision support in cardiovascular medicine. BMJ. 2022 May 25;377: e059818. DOI: 10.1136/bmj-2020-059818.
Florien S van Royen, Folkert W Asselbergs, Fernando Alfonso, et al. Five critical quality criteria for artificial intelligence-based prediction modelsJ. . European Heart Journal, 2023, ehad727, https://doi.org/10.1093/eurheartj/ehad727.
Khera R, Simon MA, Ross JS. Automation Bias and Assistive AI: Risk of Harm from AI-Driven Clinical Decision Support. JAMA. 2023,330(23):2255-2257. DOI:10.1001/jama.2023.22557.
Silcox, C., Dentzer, S., Bates, D.W. AI-enabled clinical decision support software: A “trust and value checklist” for clinicians. NEJM Catalyst. 2020,1(6). DOI: https: //doi.org/10.1056/cat.20.0212.

No competing interests reported.

SupplementalTable.docx

Download PDF

Editorial decision: Revision requested
09 Sep, 2024
Editor assigned by journal
05 Sep, 2024
Submission checks completed at journal
05 Sep, 2024
First submitted to journal
31 Aug, 2024

You are reading this latest preprint version

Current status of VTE risk assessment and prevention using clinical decision support system: a cross-sectional survey from China

Status:

Version 1

Abstract

Introduction

Materials and Methods

Results

Geographical and Demographics Characteristics

Implementation details of VTE-CDSS

Implementation status of VTE-CDSS assisted in monitoring quality indicators

The efficiency and effectiveness of using VTE-CDSS

Barriers in using VTE-CDSS

Problems to be further optimized and solved in the future in the implementation of VTE-CDSS

Discussion

Improving system functionality to adapt to clinical pathways for VTE risk assessment and prevention

Deeply understand and implement the "Five Correct" principles of VTE-CDSS

Promote the application of VTE standard datasets to improve data quality and availability

Improve the transparency and interpretability of VTE-CDSS related AI algorithms

Strengths and Potential Limitations

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1