How Artificial Intelligence could improve fracture detection in horses

Researchers at the Royal Veterinary College (RVC) are exploring how artificial intelligence (AI) could help veterinarians detect fractures in animals more quickly and accurately. Their work has been shortlisted for the STEM for Britain 2026 award and highlights how technology developed for human medicine can also benefit animal health and welfare.

Fractures are a major concern in Thoroughbred racehorses. These injuries can affect both a horse’s welfare and its racing career, and in severe cases they can be life-threatening. Studies estimate that around 10% of racehorses experience a fracture during training, while bone injuries occur in roughly 1.3 per 1,000 starts in flat racing. Because of this risk, early and accurate detection of bone damage is extremely important.

However, diagnosing fractures is not always straightforward. Veterinarians usually rely on radiographs (X-rays) to assess suspected bone injuries. While X-rays are very useful, identifying fractures on these images can be challenging. Small cracks or subtle changes in bone structure may be difficult to see, and image quality or the angle at which the X-ray is taken can also affect interpretation. As a result, there is growing interest in using technology such as AI to support clinical decision-making.

The RVC research team developed an AI system designed to analyse medical images and identify fractures. The study was led by Associate Professor of Statistics Dr Ruby Chang, with the research carried out by Dr Hanya Ahmed. To train the system, the researchers created a large database of images that included 100 equine fracture cases collected from two UK equine hospitals and from published studies. They also included 70 feline fracture cases and around 4,000 human fracture images from a public database.

The AI system works in three stages. First, it identifies the type of medical image being analysed, such as an X-ray, CT scan, or MRI scan. Next, it determines the angle or projection of the image. Finally, it analyses the image to detect whether a fracture is present and to pinpoint its exact location.

One particularly interesting feature of the study is the use of a technique called transfer learning. In transfer learning, an AI model is first trained on a large dataset (in this case, thousands of human fracture images). The knowledge it gains is then adapted to a smaller dataset from another field; in this case, veterinary medicine. Because there are far fewer veterinary medical images available for training, this approach helps overcome one of the main challenges of developing AI systems for animal healthcare.

Using this method, the AI system was able to detect and locate fractures in horses with accuracy levels between 71% and 84%, despite having a relatively small number of equine images to learn from. The system also achieved very high accuracy when identifying image types and projections, reaching more than 96% accuracy in some stages of the analysis.

The results suggest that AI could become a valuable support tool for veterinarians. By helping identify fractures more quickly and reliably, AI-assisted systems may reduce uncertainty in diagnosis and allow treatment to begin earlier. This could improve recovery outcomes for horses and other animals.

The research team is now expanding the project through collaboration with the Hong Kong Jockey Club. The next goal is to investigate whether AI can detect early bone changes before a fracture occurs. If successful, this could help prevent serious injuries in racehorses and improve welfare within the sport.

Although the current study focuses on horses, the approach could also be adapted for other species such as cats, dogs, and potentially humans. Overall, the research demonstrates how advances in AI and medical imaging could play an increasingly important role in the future of veterinary diagnostics.

 

For more details, see:

Ahmed, Hanya T., Dagmar Berner, Qianni Zhang, Kristien Verheyen, Francisco Llabres-Diaz, Vanessa G. Peter, and Yu-Mei Chang. 2026.

Bridging Species with AI: A Cross-Species Deep Learning Model for Fracture Detection and Beyond

Bioengineering 13, no. 2: 213.

https://doi.org/10.3390/bioengineering13020213

Does starting racehorses early affect their careers? Insights from a German study

 

© Nataliavo Dreamstime.com

When should racehorses begin training and competition? This is a long-standing question in equine science and
welfare. While many horses in other disciplines do not start ridden work until they are around three or four years old, Thoroughbred racehorses are often introduced to training much earlier. A recent study using data from German Thoroughbred racing has explored whether starting training and racing at a young age affects a horse’s racing career.

The research, led by Mailin Hein and colleagues at the University of Veterinary Medicine Hannover in collaboration with Dr Monica Venner from the Equine Clinic Destedt, analysed historical racing data to investigate whether early training leads to shorter or less successful careers. The team used records from the German Racehorse Association, examining the racing histories of 600 Thoroughbred horses.

Racehorses typically begin training at 18–24 months of age and may race as two-year-olds. However, horses do not reach full skeletal maturity until around six to eight years of age, meaning their bones and joints are still developing during early training. Because of this, some experts question whether starting intense work too early could increase injury risk or shorten a horse’s racing career.

The debate is particularly relevant in Germany. German animal welfare legislation generally prohibits training or competing with horses younger than 30 months, although Thoroughbred racing is currently exempt while researchers investigate the welfare implications of early training.

The researchers carried out a retrospective study, meaning they analysed existing records rather than conducting a new experiment. The 600 horses in the dataset were divided into three groups depending on when they began training and racing:

• Early training / early racing – horses that entered training between 16 and 24 months and raced as two-year-olds
• Early training / late racing – horses trained early but did not race until after the age of two
• Late training – horses that began training later, between 25 and 30 months

The researchers compared several indicators of racing success between the groups. These included career length, performance ratings, total number of races, and lifetime earnings.

One interesting finding was that sex influenced career length. Geldings (castrated male horses) tended to have longer racing careers than other horses. This is likely because stallions and mares may retire earlier to enter breeding programmes, while geldings do not have this option. Because of this difference, the researchers looked at geldings as a separate group.

Among geldings, horses that started racing at two years old achieved higher maximum performance ratings than those that first raced at three years of age or later. Importantly, the early-racing group did not show shorter racing careers compared with horses that started racing later.

Overall, the study found no evidence that early training or racing at two years old had a negative effect on career length or performance in the horses analysed.

The findings suggest that, at least in this group of German Thoroughbreds, beginning training and racing at a young age did not appear to reduce racing success or shorten careers. However, the researchers emphasise that the results should be interpreted carefully.

Because the study used historical data, it cannot fully explain why some horses started racing earlier than others. For example, horses that mature physically earlier may naturally be selected to race sooner. Horses with health problems may also leave training early and therefore not appear in racing records.

The researchers suggest that future studies should include veterinary records, training information, and management factors, such as housing conditions and workload. Understanding these factors could help scientists evaluate how a horse’s physical development, health, and mental wellbeing influence its ability to cope with early training.

Ultimately, research like this aims to improve the welfare and management of racehorses, helping trainers and veterinarians make evidence-based decisions about when young horses should begin their athletic careers.

 

For more details, see: 

 

Hein, Mailin, Nina Volkmann, Jeanette Probst, Nicole Kemper, and Monica Venner. 2026.

Thoroughbred Geldings′ Career: Influence of Age at the Start of Training and Racing

Animals 16, no. 4: 576.

https://doi.org/10.3390/ani16040576

Free articles on equine infectious disease and Microbiology in EVJ Special Collection

© Golop Dreamstime.com

The March 2026 issue of Equine Veterinary Journal (EVJ) highlights major advances in equine infectious disease and microbiology, with a Special Collection dedicated to the equine microbiome. The issue, which is free to access for 12 weeks, features 30 research papers spanning a wide range of diseases, body systems, and pathogens. Many of the studies involve international collaboration, reflecting the global importance of managing infectious diseases in horses.

Equine infectious disease remains a key research focus for veterinarians and scientists because of its impact on horse health, welfare, and the equine industry. The special issue identifies three major themes: the importance of national disease surveillance programmes, the need for accurate and definitive diagnoses, and the value of comparing diseases across different animal species to better understand how pathogens behave.

Several studies focus on improvements in diagnostic techniques, particularly for equine fever, equine influenza, and gastrointestinal parasites. These advances aim to support faster detection of disease and enable earlier implementation of biosecurity measures, which can help reduce transmission between horses.

Research into gastrointestinal diseases also features prominently. Studies explore the use of metabolic profiling to detect equine proliferative disease earlier in foals, investigate possible causes of colitis, and emphasise the importance of additional diagnostic testing for chronic hepatitis and liver failure in cases linked to hepacivirus infection. Together, these findings may support earlier intervention and improved disease management.

Antimicrobial resistance is another significant area of concern. One survey found that nearly 40% of hospitalised horses received at least one antimicrobial treatment. Research linked to colic surgery suggests that using antimicrobials for more than 24 hours after surgery is unnecessary for preventing complications, highlighting opportunities to reduce antibiotic use and slow the development of antimicrobial resistance.

Several papers examine diseases affecting multiple body systems. For example, updated guidance on equine botulism emphasises its importance as a differential diagnosis when horses show neurological signs. Other studies investigate multisystem diseases by comparing similar infections in other species, helping researchers better understand how diseases develop and present in horses.

Blood-related infectious diseases are also discussed. Two systematic reviews examine equine trypanosomiasis, focusing on its prevalence, management, and control strategies. Another study introduces a new diagnostic tool for detecting Theileria haneyi, a parasite associated with equine piroplasmosis, and reports the first evidence of its presence in China.

Regionally important diseases are highlighted through research on Hendra virus, which uses government and field data to study how domestic horses may act as bridging hosts for zoonotic pathogens. Another study shows that horses infected with African Horse Sickness virus can develop severe blood clotting disorders that cause significant bleeding. Research also suggests that equine encephalosis virus should be considered when diagnosing African Horse Sickness in regions where both diseases occur.

The issue also examines reproductive health, with one study linking Escherichia coli infection to reduced fertility in Thoroughbred broodmares. Another uses multi-omics technology to study nocardioform placentitis, proposing that the condition be renamed mucoid placentitis due to the variety of bacteria involved.

Respiratory disease research forms a major section, including several studies on strangles. Findings emphasise the importance of screening, quarantine, monitoring, and disease reporting to improve prevention. Additional studies investigate respiratory disease caused by Rhodococcus equi, equine influenza vaccination strategies, and the use of air sampling in stables as a non-invasive surveillance method for equine herpesvirus.

Other research explores infectious skin diseases and rare ulcerative conditions, demonstrating the importance of obtaining accurate diagnoses and comparing disease patterns across species when treatment data in horses is limited.

The Special Focus section also includes a dedicated collection on the equine microbiome, updating earlier research published in 2019. This collection highlights rapid progress in understanding the complex communities of microorganisms that influence equine health.

Overall, the research presented in this EVJ issue aims to improve disease detection, diagnosis, treatment, and prevention. These advances are expected to strengthen equine welfare globally while supporting the livelihoods of communities that depend on horses for work and transport.

The Special Focus Issue is available here:

 https://beva.onlinelibrary.wiley.com/toc/20423306/2026/58/2

Identifying pain in horses faces

© Briana Hunter Dreamstime.com

  

Research led by scientists at Anglia Ruskin University, including its specialist land-based
campus ARU Writtle, has provided new insight into how well humans can recognise pain in horses. The study, published in the academic journal Anthrozoös, highlights important welfare implications and helps explain why recognising equine discomfort can be challenging, particularly for people new to working with horses.

Horses are widely used by humans for sport, leisure, and work. Because they are prey animals, horses have evolved to mask signs of pain or weakness to avoid attracting predators. While this survival strategy helps them in the wild, it can make it harder for humans to detect when a horse is suffering, which may delay treatment and negatively affect welfare.

The research team, which also included collaborators from Bournemouth University and University of São Paulo, investigated how accurately people can recognise pain by studying facial expressions. A total of 100 participants were recruited, including 30 individuals with experience caring for horses and 70 with little or no horse-care experience. Participants were shown 60 photographs: 30 of human faces and 30 of horse faces. They were asked to rate them for signs of pain, emotional intensity (arousal), and whether the expression appeared positive or negative (valence). To provide a benchmark for accuracy, 10 equine behaviour professionals also evaluated the horse images.

The results showed that people were generally better at recognising pain in human faces than in horse faces. However, participants with horse-care experience were significantly more accurate at identifying pain in horses compared to those without experience. The study also found that the number of years spent working with horses improved accuracy, suggesting that recognising equine facial expressions is a skill that develops through observation and practice.

Experienced participants were more likely to recognise subtle indicators of pain, such as changes in ear position, the angle of the eyes, and increased muscle tension around the face. These indicators form part of recognised equine pain assessment methods and are essential for early identification of discomfort.

The research also explored how human psychological traits affect the ability to recognise pain. Participants completed assessments measuring empathy and social anxiety. Interestingly, empathy levels did not appear to influence pain recognition accuracy for either humans or horses. However, social anxiety produced more complex results. People with higher social anxiety were better at recognising pain in human faces but not in horse faces. In fact, these individuals tended to overestimate pain in horses, suggesting that human emotional processing may sometimes interfere with interpreting equine body language.

According to lead researcher Nicola J. Gregory, the study provides the first scientific evidence that humans often struggle to identify pain in horses, but experience significantly improves this ability. Meanwhile, equine behaviour specialist Rosa Verwijs emphasised that horses’ natural tendency to hide pain means problems may only become noticeable when behaviour escalates, which can increase risks to both horse welfare and human safety.

Overall, the findings highlight the importance of education and practical training for anyone working with horses. Improving knowledge of equine facial expressions and behavioural indicators of pain could help owners and professionals identify problems earlier, seek veterinary treatment more quickly, and ultimately improve the welfare of horses in human care.

 

For more detail, see:

Gregory, N. J., Trimmer, M., Dempsey, T., Verwijs, R., Lencioni, G. C., & Moseley, R. L. (2026).

Reading Pain in Horse and Human Faces: The Influence of Horse Experience, Social Anxiety, and Empathy. Anthrozoös, (2026) 39(1), 161–178. https://doi.org/10.1080/08927936.2025.2551433

Pituitary Pars Intermedia Dysfunction (PPID) and the potential role of β-endorphin in diagnosis

Pituitary Pars Intermedia Dysfunction (PPID), commonly known as equine Cushing’s disease, is a common endocrine disorder affecting older horses. The condition is associated with abnormal activity of specialised cells called melanotropes, which are located in the pars intermedia region of the pituitary gland. These cells normally produce several hormones derived from a precursor molecule known as pro-opiomelanocortin (POMC). In horses affected by PPID, melanotropes become overactive, leading to excessive production of several hormones that contribute to the clinical signs observed in the disease.

One of the most important hormones produced by melanotropes is adrenocorticotrophic hormone (ACTH). This hormone stimulates cortisol release from the adrenal glands and is widely used as a diagnostic marker for PPID. Elevated ACTH concentrations in blood samples often support a diagnosis of the disease. However, ACTH may not provide a complete understanding of PPID, as melanotropes also produce other biologically active substances that may influence disease development and clinical signs.

These additional substances include alpha-melanocyte stimulating hormone (α-MSH), corticotropin-like intermediate peptide (CLIP), and β-endorphin. β-endorphin is a peptide hormone composed of 31 amino acids and functions as part of the body’s natural pain and stress regulation system. The amino acid sequence of equine β-endorphin is very similar to that found in humans, differing by only three amino acids. Because β-endorphin is also derived from POMC and secreted by melanotropes, researchers have suggested it may play an important role in the development and progression of PPID.

Traditionally, β-endorphin concentrations in horses have been measured using radioimmunoassay (RIA). However, this technique involves the use of radioactive materials, which presents safety risks and has become less widely available in recent years. In human medicine, enzyme-linked immunosorbent assays (ELISAs) are commonly used as a safer and more accessible alternative for hormone measurement. Researchers have therefore investigated whether commercially available human β-endorphin ELISA tests could be used reliably for equine samples.

A study conducted by Nathalie Fouché and colleagues at the University of Bern, Switzerland, aimed to validate a human β-endorphin ELISA kit for use in horses and to compare β-endorphin concentrations between horses diagnosed with PPID and healthy control horses. Validation of the test involved comparing standard curves generated using both synthetic equine β-endorphin and human β-endorphin. The results demonstrated full parallelism between the curves, indicating the test could accurately measure equine hormone levels.

The researchers also assessed the reliability of the assay by calculating intra-assay and inter-assay variation. These tests measure consistency within a single test plate and between multiple plates. The assay showed acceptable levels of variation, suggesting it is suitable for research use. Additionally, β-endorphin concentrations remained stable in plasma samples over a 24-hour period regardless of centrifugation timing, storage temperature, or storage duration, which supports the practicality of sample handling.

The pilot study compared five horses diagnosed with PPID to twenty healthy aged control horses. The findings revealed significantly higher β-endorphin concentrations in horses with PPID, with median concentrations of 506 pg/mL compared to 35 pg/mL in healthy horses. These results suggest that β-endorphin may be elevated in horses with PPID and could potentially provide additional diagnostic or pathophysiological information alongside ACTH testing.

Overall, this research highlights the possible importance of β-endorphin in understanding PPID and suggests that ELISA-based testing may provide a safer and more accessible method for future investigation. Further studies with larger sample sizes are required to confirm the diagnostic value of β-endorphin in horses with PPID.

For more details, see: 

N. Fouché, J. Howard, V. Gerber, P. Billmann, M. Farinha do Sul, G. Christen, R. Bruckmaier, C. Philipona, N. Besuchet Schmutz, J. Gross,

Pilot study of β-endorphin concentrations in horses with pituitary pars intermedia dysfunction using a newly validated enzyme-linked immunosorbent assay,

Domestic Animal Endocrinology (2026) vol 95,106982

https://doi.org/10.1016/j.domaniend.2025.106982

Gastrointestinal parasites in Slovak horses

© Jozef Mikat | Dreamstime.com

Gastrointestinal parasitism remains a significant health concern in equine populations worldwide. Horses commonly host a wide range of intestinal helminths (parasitic worms), some of which can cause serious disease, reduced performance, and compromised welfare.

In recent years, growing resistance to anthelmintic (deworming) medications has become a major challenge for effective parasite control. This resistance has largely developed due to management practices such as frequent blanket deworming, incorrect dosing, and repeated use of the same drug classes. As a result, current parasite control strategies are shifting toward evidence-based approaches that rely on monitoring infection levels and targeting treatment appropriately.

Slovakia has a well-established horse breeding industry, with over 22,500 horses representing approximately 40 breeds.

To support improved management strategies, recent research conducted by Kuzmina and colleagues aimed to evaluate the distribution and prevalence of gastrointestinal helminths in horses in eastern Slovakia, while also examining management factors that may influence infection levels.

The study involved analysis of 392 faecal samples collected from horses across 24 farms, representing a range of ages, breeds, and management systems. Coprological diagnostic techniques were used to identify parasite eggs. The McMaster method, with a sensitivity of 50 eggs per gram (EPG), was employed to detect nematode infections, while a double-centrifugation sedimentation-flotation method was used to identify tapeworm (Anoplocephala spp.) eggs. Additional information regarding horse age, breed, stocking density, and parasite control practices was collected to allow assessment of risk factors associated with infection.

Results indicated that strongylid nematodes were the most prevalent parasites identified, with 65.6% of horses testing positive. Egg counts ranged widely from 50 to 2800 EPG, demonstrating substantial variation in parasite burdens between individual animals. Notably, approximately 29.8% of horses were responsible for 80% of the total strongylid egg output, supporting previous evidence that parasite burdens are typically unevenly distributed within equine populations. This finding reinforces the principle behind targeted selective treatment, where only horses with higher egg counts are treated to reduce drug resistance and maintain refugia (untreated parasite populations).

Parascaris spp., which primarily affect younger horses, were detected in 4.5% of samples. Tapeworm infections caused by Anoplocephala spp. were relatively uncommon, being detected in only 0.3% of horses, and no additional helminth species were identified. These results confirm that strongylids and Parascaris spp. remain the dominant equine gastrointestinal parasites in Slovakia.

Statistical analysis revealed that horse age and stocking density were significant predictors of strongylid infection levels. Younger horses and animals kept at higher stocking densities were more likely to exhibit higher parasite burdens. In contrast, infection with Parascaris spp. was influenced solely by horse age, with younger horses demonstrating greater susceptibility due to their developing immune systems.

Overall, this study provides valuable insight into the current epidemiology of equine gastrointestinal parasites in Slovakia. The findings highlight the importance of regular faecal egg count monitoring and demonstrate that parasite burdens vary significantly between individuals. Incorporating targeted selective treatment strategies, alongside improved pasture management and reduced stocking density, may help slow the progression of anthelmintic resistance.

 

For more details, see:

Tetiana A. Kuzmina, Alžbeta Königová, Ludmila Burcáková, Yaroslav Syrota, Michal Babják, Marián Várady,

Gastrointestinal parasite occurrence in Slovak horses and factors affecting Strongylidae and Parascaris spp. egg shedding,

Veterinary Parasitology: Regional Studies and Reports,(2025)  Vol 64, 101328,

ISSN 2405-9390,

https://doi.org/10.1016/j.vprsr.2025.101328