The small ruminant farming industry has seen steady expansion in the past few decades and has become a crucial source of employment and income in family agricultural areas (Hostiou et al., 2020). One of the main obstacles limiting the productivity of sheep and goats is infection with gastrointestinal parasites (GIP), resulting in substantial economic losses through diminished weight gain and concomitant impaired production of meat, wool, and milk (Starkey and Pugh, 2020; Arsenopoulos et al., 2021). Furthermore, these parasites constitute significant health hazards, frequently leading to production losses and even mortality of young animals and their mothers during and following parturition through weaning (Torres-Acosta et al., 2012).
Haemonchus contortus, an ubiquitous blood-feeding GIP of small ruminants world-wide (Bath et al., 2011; Kuiseu et al., 2021), due to its high pathogenicity, including anemia and hypoproteinicity, can significantly decrease the health and production of infected animals (Charlier et al., 2020, Fernandes et al., 2022). Effectively managing this parasite is essential for maintaining the health and economic sustainability of the small ruminant livestock industry (Charlier et al., 2020). Anemia presents substantial health obstacles on a global scale, affecting both human health and the well-being and efficiency of large and small ruminants, especially in areas with limited access to veterinary services (Biffa et al., 2006; Constable et al., 2016). Furthermore, livestock populations in these underserved areas often experience nutritional inadequacies, and together with a variety of parasitic infections, these additional factors can also have a significant impact on the occurrence of anemia (Panda et al., 2004). Anemic animals can be saved, requiring interventions such as providing additional nutrients and deworming treatments, but if not treated on time, this condition may lead to death of infected animals. Unfortunately, these interventions are often not available in underserved areas with limited resource farmers.
In light of the above, anemia needs to be addressed in livestock, especially small ruminants, particularly in agricultural areas where these animals are a primary source of nutrition and family income (Pulina et al., 2017). Simple, rapid diagnostic testing for anemia is crucial, as it provides an objective and quantitative measure of health that goes beyond the constraints of subjective symptom evaluation (Fitzpatrick, 2013).
Two diagnostic tests currently in use are measuring an animal’s HCT or using the FAMACHA system of anemia detection (van Wyk and Bath, 2002). The HCT level, which measures the percentage of blood volume occupied by red blood cells (RBCs), is a widely used diagnostic indicator (Asaduzzaman et al., 2018), but is not practical for producers, as it requires use of anti-coagulant blood tubes (with blood collected by venipuncture) and specialized laboratory equipment (microhematocrit blood tubes and centrifuge, etc.). Researchers have used this procedure to validate the FAMACHA score chart, which is used to match colors on a laminated card to the color of the lower eyelid conjunctiva of sheep or goats with varying levels of anemia (van Wyk and Bath, 2002; Kaplan et al., 2004). For example, an HCT score above 23% corresponds to FAMACHA scores of 1 or 2 and indicates healthy animals (Yilmaz et al., 2016). An HCT value between 17% to 23% is considered borderline, corresponding to a FAMACHA score of 3 (treatment may or may not be given based on farmers’ decision), while an HCT value between 12% to 17% is considered as anemic, with a FAMACHA score of 4. An HCT value below 12% indicates severe anemia in small ruminants and corresponds to a FAMACHA score of 5 (Kaplan et al., 2004). Both the HCT procedure and the FAMACHA system have been widely adopted worldwide by researchers and farmers, respectively, but both also can be time-consuming and require specialized training for proper use. In addition, matching of color of the eye conjunctiva of sheep and goats can be greatly influenced by environmental conditions (sunny or cloudy day) or person to person differences, leading to false negative and false positive diagnoses (Kaplan et al., 2004). As an alternative, recent progress has led to the creation of portable, user-friendly instruments that enable swift HCT evaluation, even by lab workers, including students and researchers with limited training, although most of these were developed for point-of-care in human medicine.
There have been significant advancements in optical techniques for measuring HCT in human subjects, particularly using microfluidic settings (Jalal et al., 2017; Kang, 2018), which allow for direct correlation between the greyscale intensity changes of blood and its HCT values. Nevertheless, these procedures frequently encounter issues due to their susceptibility to changes in ambient lighting, which can lead to distorted outcomes in different operational scenarios. However, recent advancements in the detection of HCT on silicon chips employing impedimetric techniques have demonstrated the potential use of red blood cell suspensions in phosphate-buffered saline instead of whole blood, thereby excluding substantial plasma conductivity influences that could potentially impact precision (Kuan & Huang, 2020). The complexity of this technology also presents issues for point-of-care (POC) applications (Chakraborty et al., 2020; Treo et al., 2005). Similarly, centrifugal microfluidics have been used to determine HCT levels and complete blood counts on a motorized plastic disk (Pishbin et al. 2015)), but this approach also has constraints in POC environments due to its reliance on uninterrupted power provision and challenges associated with the disposal of nonbiodegradable waste (Agarwal et al., 2020; Riegger et al., 2007).
Functionalized paper strips have emerged as viable platforms for efficient medical diagnostics using blood samples (Gilmore et al., 2016; Komatsu et al., 2020), and Berry et al. (2016) created a paper microfluidic system that combined vertical and lateral channels to regulate the movement of blood cells. Although this latter procedure is simple, it lacks quantitative precision when compared to traditional laboratory tests and necessitates specific manufacturing processes (Berry et al., 2016).
In summary, the evaluation of blood hemoglobin levels continues to be a vital method for detecting anemia (Williams et al. 2023). There are several miniature platforms in support, such as paper-based devices that are currently being developed for this end, such as one from Laha et al. (2022), comprising a hemoglobin sensor that is integrated into a smartphone via colorimetric signals detected on a paper strip. While being cost-effective and accurate, this method requires meticulous sample preparation and careful handling of reagents. Additionally, the necessary chemical reagents frequently require refrigeration, which may be delimiting, particularly in limited resource farming areas. Frantz et al. (2020) also described a fast and cost-effective method employing a smartphone and a non-reactive lateral flow device to determine HCT levels. Nevertheless, it necessitates the use of high-resolution videography to follow blood flow, which can be expected to be impractical in low-resource environments, but also elsewhere, due to the requirement for specialized equipment and experienced staff.
An assessment of the current POC technologies for HCT determination emphasizes notable difficulties, such as the requirement for intricate strip production, the absence of dependable supply chains for sensitive reagents, and the inadequacy of advanced detection technologies for extreme POC circumstances. In the present investigation, the process, from data collection to analysis of images, which has been modified and compared using four different artificial intelligence-machine learning based models that were developed by Laha et al. (2022), is the use of an inexpensive sensor for determining HCT levels using filter paper strip and smartphone-based image analytics .
This novel method utilizes the basic principles of viscous fingering, in which a less viscous fluid (blood) displaces a more viscous fluid (glycerol), resulting in different interfacial patterns (Laha et al. 2022). Through the examination of these patterns, the Hausdorff fractal dimensions (also known as fractal dimension analysis) can be calculated to establish a correlation with HCT levels (Laha et al. 2022). The above technique necessitates only a single image of the blood pattern, obtained using a smartphone camera, greatly streamlining the procedure. This method is suited for rapid, on-site testing by individuals with minimal training since it avoids the requirement for specialist reagents or fabrication stages. It involves use of filter paper soaked in glycerol, an innovation with the capacity to transform the process of identifying anemia in locations with limited access to healthcare, in line with global health goals, by enabling precise and immediate identification of anemia. Point-of-care devices are transforming the management of anemia in humans by allowing timely interventions that can greatly enhance health and output, but there has been very limited work to date on rapid identification of anemia level or development of rapid anemia identification biosensors in domestic or farm animals used for production.
Therefore, the research goal in this investigation was to develop an AI-ML supported remote real-time livestock health monitoring system as a tool to assist veterinarians, livestock farmers and other stakeholders for confirming anemia levels in animals. The specific objectives to achieve this goal were as follows:
- Develop an easy-to-use and efficient image acquisition mechanism to obtain quality data from goats for real-time analyses.
- Compare the various AI-ML based models to train, test, and validate acquired data for determining animal health status through an optimization approach.
- Develop a real-time remote animal health monitoring system using the best optimized model.