The appeal of using dogs to study brain aging is not sentimental; it is scientific. Dogs share our homes, our air, and in many cases our couches, which means they are exposed to many of the same environmental influences that shape human health. They also develop age-related neuropathology that resembles aspects of human dementia, including the accumulation of amyloid-related pathology and declines in hippocampal and prefrontal function. Yet unlike laboratory rodents, dogs live rich, variable lives, and unlike humans, they age on a compressed timescale of years rather than decades. A seven-year-old dog is already entering the window in which subtle cognitive differences begin to emerge, making the species uniquely positioned for longitudinal studies that would take a lifetime to complete in people.

The problem, until now, has been measurement. Existing canine cognitive assessments tend to rely heavily on owner questionnaires, which capture noticeable dysfunction but often only after decline is well established. Others are laboratory-based batteries that require specialized equipment, extensive animal training, or repeated visits that community dogs and their owners cannot easily accommodate. CANID was designed to thread this needle: a functional evaluation consisting of three cognitive tasks that probe memory and executive function, the very domains that erode earliest in human aging, deliverable in a single session of roughly forty minutes, and suitable for dogs living in ordinary homes rather than research colonies.

The technical architecture of the battery reflects decades of work in human neuropsychology. Executive function, the umbrella term for the cognitive control processes mediated largely by prefrontal cortex, is assessed through tasks requiring behavioral inhibition, attention, and problem-solving under changing rules. Memory components tap working memory and the ability to hold and manipulate information across short delays. Each task yields a subscore, and the subscores combine into an overall CANID score, giving researchers both a sensitive global index and finer-grained domain-specific measures. This structure mirrors the logic of human test batteries, in which a single composite score can track broad decline while individual subscores localize which cognitive systems are faltering.

Validation is where most new behavioral measures stumble, and where CANID makes its strongest case. The team developed their scoring system using a baseline sample of 239 dogs, then brought the same animals back for repeat testing at six months, with 138 dogs returning, and again at twelve months, with 114 dogs completing the full arc. Repeatability, the statistical property that a measure returns similar values when the underlying trait has not changed, was quantified using intraclass correlation coefficients, or ICCs. The overall CANID score showed substantial repeatability at six months, with an ICC of 0.63 and a 95 percent confidence interval spanning 0.51 to 0.72, and remained substantial at twelve months, with an ICC of 0.59 and a confidence interval of 0.47 to 0.70. Critically, every task-specific subscore also demonstrated substantial repeatability, indicating that the individual cognitive measures, not just the composite, are stable enough for longitudinal research.

Those numbers deserve unpacking for anyone who has not spent time with psychometrics. An ICC of 0.63 means that roughly 63 percent of the variance in test scores at follow-up can be attributed to stable differences between individual dogs rather than measurement noise or day-to-day fluctuation. In behavioral research, where a single anxious morning or an interesting smell outside the testing window can derail a session, that level of stability is genuinely difficult to achieve. It means a researcher can administer CANID today, administer it again in six months, and have reasonable confidence that any change in score reflects change in the dog rather than change in the instrument. Without that property, longitudinal studies of cognitive aging would be chasing statistical ghosts.

The sensitivity findings may prove even more consequential. CANID detected age-related cognitive differences beginning in mid-life, well before the point at which owners would typically report concerns or questionnaires would flag dysfunction. This matters because the scientific consensus in human Alzheimer’s research has shifted decisively toward early detection: by the time clinical symptoms appear, the underlying pathology has been accumulating for years or decades. A canine measure that registers subtle decline years before overt impairment gives researchers a much wider window in which to study the trajectory of aging, test interventions, and identify predictors of who will decline fastest. In the compressed canine lifespan, that window is measured in precious months and years.

One of the study’s more honest findings concerns practice effects, the improvement in test scores that comes simply from having taken a test before. At the six-month follow-up, dogs showed a modest practice effect of 0.23 standard deviations improvement in overall score relative to baseline, a magnitude the authors note is similar to what has been reported in human studies. Intriguingly, no evidence for practice effects emerged at the one-year follow-up, suggesting the familiarity advantage fades with time. For researchers, this is both a warning and a gift: it means six-month comparisons should account for the expected bump, but it also means the effect is predictable, quantified, and consistent with the human literature, making it manageable rather than fatal to longitudinal designs.

The team has also made an unusually strong commitment to open science. To facilitate adoption, they have released detailed protocols, video libraries, data recording instruments, experimenter training guidelines, and R code for scoring in an Open Science Framework repository. This is not a trivial gesture. Behavioral measures live or die on standardization, and the history of comparative cognition is littered with tasks that produced inconsistent results because each laboratory implemented them slightly differently. By providing video libraries that show exactly how testing should look and code that automates scoring, the authors are attempting to ensure that a CANID score from one lab means the same thing as a CANID score from another, a prerequisite for the large multi-site studies that cognitive aging research increasingly demands.

The work was carried out within the framework of the Dog Aging Project, a large-scale longitudinal study of companion dogs supported in part by the National Institute on Aging, with additional support from the Glenn Foundation for Medical Research, the Tiny Foundation Fund at Myriad Canada, the WoodNext Foundation, and the Dog Aging Institute. All procedures were approved by the University of Arizona Institutional Animal Care and Use Committee, and testing was designed with welfare in mind: data collection was discontinued if dogs became fearful or anxious and could not be soothed, and no testing occurred with dogs who could not acclimate within roughly fifteen minutes of arrival. Written consent was obtained from all owners.

What comes next could be transformative for both veterinary and human medicine. With a validated, efficient, and repeatable measure in hand, researchers can begin tracking thousands of dogs across their lifespans, correlating cognitive trajectories with genetics, diet, exercise, and eventual neuropathological findings. Dogs could serve as a bridge species in which candidate interventions for human dementia are tested under ecologically realistic conditions, and veterinarians could gain a practical tool for identifying cognitive decline in clinical practice years before it becomes obvious to owners. For the millions of aging dogs sharing homes around the world, and for the humans who love them, a forty-minute test may quietly change what it means to grow old together.

Subject of Research: A validated cognitive test battery for detecting age-related neuropsychological impairment in dogs

Article Title: Cognitive assessment for neuropsychological impairments in dogs (CANID): an efficient, repeatable, and sensitive evaluation for studies of cognitive aging

Article References: Cognitive assessment for neuropsychological impairments in dogs (CANID): an efficient, repeatable, and sensitive evaluation for studies of cognitive aging. (n.d.). https://doi.org/10.1007/s10071-026-02112-x

Image Credits: AI Generated

DOI: 10.1007/s10071-026-02112-x

Keywords: cognitive aging, dogs, CANID, neuropsychological assessment, executive function, memory, Dog Aging Project, animal cognition, repeatability, dementia, open science, veterinary science

Reproductive performance is the beating heart of dairy profitability. A cow that conceives quickly after calving maintains a short calving interval, spends more of her life producing milk at peak levels, and costs her owner less in feed, labor, and veterinary care. A cow that lingers open, by contrast, drains resources month after month. Researchers have long known that a constellation of factors influences this interval, including the age of the animal and a suite of postpartum diseases such as metritis, mastitis, ketosis, retained placenta, and ovarian cysts. Epidemiological studies stretching back decades have documented, for example, that metritis reduces conception rates and prolongs the interval to subsequent pregnancy, while subclinical ketosis delays the return of estrus and extends the time from calving to both first insemination and conception.

The problem with conventional machine learning in this domain has been trust. Powerful algorithms such as gradient boosting ensembles and deep neural networks can capture nonlinear relationships among predictors that traditional regression models miss, but they do so inside a black box. A veterinarian handed a prediction has no way of knowing whether the model is responding to a genuine biological signal or to some artifact of the data. This tension between accuracy and interpretability has fueled the rise of explainable artificial intelligence, or XAI, a family of techniques designed to pry open the decision-making machinery of complex models. Among these techniques, the most theoretically robust is SHAP, short for SHapley Additive exPlanations, which borrows its mathematics from cooperative game theory.

SHAP treats every input variable as a player in a game where the payout is the model’s prediction. By systematically evaluating how the prediction changes as each feature is added or removed across all possible coalitions of features, SHAP assigns each variable a value quantifying its marginal contribution. For an individual cow, this means the method can say precisely how much her age, her history of mastitis, or her bout with ketosis pushed the predicted interval to first insemination up or down relative to the average. Aggregated across the whole dataset, these local explanations yield a global ranking of feature importance. Crucially, the method satisfies axioms of local accuracy, consistency, and missingness that competing explanation techniques such as partial dependence plots and LIME do not simultaneously guarantee, which is why it has become the interpretability tool of choice in animal science.

In the new study, a team led by Elif Çelik Gürbulak of Erciyes University assembled 466 observational records from a dairy cattle farm, later refined to 439 complete cases after excluding heifers and records with missing values. The outcome variable was the number of days from calving to first insemination, with shorter intervals signaling better reproductive performance. The predictors were age and binary indicators for five common postpartum conditions. The researchers fitted an eXtreme Gradient Boosting, or XGBoost, regression model, tuning its hyperparameters through a grid search combined with five-fold cross-validation across 432 parameter combinations. The optimal configuration used a maximum tree depth of two, a learning rate of 0.05, and subsampling rates of 0.7, with 100 boosting rounds. Interpretability was then layered on top using the TreeSHAP algorithm, which computes exact Shapley values for tree-based models in polynomial rather than exponential time.

The global picture that emerged from the SHAP analysis was strikingly coherent with veterinary intuition. Age dominated the model, posting the highest mean absolute SHAP value of 3.205 and accounting for nearly ninety percent of the model’s gain in the XGBoost importance metrics. Older cows contributed strongly to predictions of longer intervals to first insemination, a pattern consistent with earlier findings that conception probability declines with increasing parity and that pregnancy rates fall after the fourth lactation. Mastitis ranked second with a mean absolute SHAP value of 0.324, followed by retained placenta at 0.173, ovarian cysts at 0.079, ketosis at 0.073, and metritis at 0.072. The positive SHAP contributions of the postpartum diseases indicated that their presence pushed predictions toward longer intervals, aligning with known biology in which mastitis pathogens trigger prostaglandin release, disrupt the estrous cycle, and impair oocyte quality and early embryonic development.

When the researchers stratified the herd into low, medium, and high reproductive performance classes using percentile thresholds, the disease burden told a clear story. Mastitis prevalence climbed progressively from 19.6 percent in the high-performance group to 38.4 percent in the low-performance group, while retained placenta affected 21.2 percent of the lowest performers. The mean interval to first insemination stretched from 56.27 days in the top group to 78.47 days in the bottom group. Radar charts visualizing the ten best and ten worst animals reinforced the gradient, showing that higher overall disease prevalence accompanied poorer reproductive outcomes. At the level of individual cows, the SHAP values varied in both direction and magnitude, revealing heterogeneity that a single herd-level statistic would have concealed entirely.

Yet the study is equally notable for what it honestly admits it cannot do. On the independent test dataset, the model’s predictive performance collapsed: the root mean square error rose to 39.05 days, the mean absolute error to 18.58 days, and the coefficient of determination fell to a mere 0.01, meaning the model explained essentially none of the variance in unseen animals. The training set metrics, with an R-squared of 0.18, were only modestly better. The authors are explicit that the SHAP importance values should be read as explanations of the patterns the fitted model learned, not as definitive evidence of causal or universally generalizable biological relationships. Important variables influencing reproduction, such as nutrition, genetics, environment, and management practices, were either absent from the dataset or insufficiently represented, and the modest sample size of a single farm further limits external validity.

This candor is precisely what makes the study a useful template rather than an overhyped failure. The authors frame the work as a methodological proof of concept, illustrating how XGBoost and SHAP can be integrated into veterinary data analysis before any clinical decision-support application is attempted. Their conclusion echoes a growing consensus in the explainable AI literature: a model should be judged not only by its predictive accuracy but by the quality and trustworthiness of its explanations, and transparent attribution of feature contributions is a prerequisite for any tool a veterinarian might actually rely on at the bedside, or in this case at the stall side. Future iterations, the researchers argue, will need larger and more diverse herds, a broader range of reproductive, nutritional, genetic, and environmental variables, and validation across multiple farms before such models can inform real herd-management decisions.

The broader significance extends well beyond dairy science. As machine learning spreads through livestock production, companion animal medicine, and epidemiology, the demand for models that can justify their predictions will only intensify. Studies applying SHAP have already illuminated infection risk in cats, mortality prediction from veterinary health records, and survival in equine colic, and comparisons with traditional logistic regression continue to clarify where each approach belongs: regression for causal interpretation, SHAP-augmented machine learning for clinical interpretation of complex patterns. The Turkish team’s dairy cow study, with its transparent accounting of both strengths and limitations, offers a candid glimpse of that future, one in which the algorithms advising farmers and veterinarians can finally show their work.

Subject of Research: Explainable machine learning assessment of reproductive performance and disease risk factors in dairy cows

Article Title: Assessment of reproductive performance in dairy cows using explainable machine learning

Article References: Çelik Gürbulak, E., Kara, U., Canooğlu, E., Yüceel, H. A., Çetin, E., Demirel, M., & Gürbulak, K. (2026). Assessment of reproductive performance in dairy cows using explainable machine learning. Archives Animal Breeding, 69(3), 469-476. https://doi.org/10.5194/aab-69-469-2026

Image Credits: AI Generated

DOI: 10.5194/aab-69-469-2026

Keywords: dairy cows, reproductive performance, explainable artificial intelligence, XGBoost, SHAP, machine learning, veterinary science, mastitis, ketosis, retained placenta, metritis, ovarian cysts

The scale of the problem is enormous. The American Veterinary Medical Association estimates that more than 4.5 million people in the United States are bitten by dogs every year, and the Centers for Disease Control and Prevention reports that roughly 20 percent of those bites require medical attention. Beyond the human toll, the economic burden is substantial: a 1999 estimate placed the annual cost of medical care for dog bite victims between 235.6 and 253.7 million dollars, a figure that rises to between roughly 471.5 and 507.7 million dollars when adjusted to 2026 dollars. And victims are not only human. Dogs bite other dogs, and cats — because of their small size — can suffer devastating injuries from the same bite that would leave a human moderately hurt.

The researchers drew their data from the Humane Society of the Pikes Peak Region (HSPPR), the agency responsible for animal law enforcement in El Paso and Pueblo counties, home to Colorado Springs and Pueblo. HSPPR responds to around 2,300 calls each year about aggressive companion animals and publishes reported bites on a public “Bite Blotter.” Each victim in an incident receives a separate record, and HSPPR classifies severity in three tiers: moderate, meaning one deep puncture or multiple moderately deep punctures and lacerations; severe, meaning multiple punctures, deep gashes, serious bodily injury, or permanent disfigurement; and fatal, meaning the death of a human or domestic animal. The study region is also demographically diverse, with El Paso County’s 757,040 residents contrasting with Pueblo County’s 169,277, roughly half the bachelor’s degree attainment and double the poverty rate, which the authors argue improves the generalizability of their findings.

The analytical approach was deliberately more nuanced than most prior work. Instead of treating bite outcomes as a simple yes-or-no event, the researchers used ordered logistic regression on the full sample of human, dog, and cat victims, with severity as a three-level ordinal outcome. They tested the sex of the biting dog, whether the dog was spayed or neutered, an interaction between those two variables, victim species, and the location of the bite on the victim’s body. A Brant test confirmed the proportional odds assumption held for the model. For the human-only subsample, where no bites proved fatal, severity became a binary outcome of moderate versus severe, and the team turned to logistic regression, adding victim age and a quadratic age term to probe for nonlinear effects.

The first major result was a surprise. Across the full sample, neither the sex of the biting dog nor its alteration status showed any statistically significant association with bite severity, and the models’ near-zero pseudo R-squared values confirmed that these variables were poor predictors of how bad a bite would be. What did matter enormously was the species of the victim. Compared with cat victims, dog victims had 92.8 percent lower odds of a more severe bite (odds ratio 0.072), and human victims fared even better, with odds 97.9 percent lower (odds ratio 0.021). The authors caution that reporting bias likely inflates this gap: a moderate bite to a cat may simply go unnoticed or unreported, while the same injury to a person almost certainly reaches animal control. Bite location also mattered — injuries to the torso or to multiple areas of the body were roughly three times more likely to be severe than bites to the hands, paws, arms, or front legs.

When the analysis narrowed to human victims only, a clearer picture emerged. Bites from male dogs carried 81.5 percent higher odds of being severe rather than moderate compared with bites from female dogs (odds ratio 1.815), an association that held consistently across every model specification. Intriguingly, whether the dog was neutered made no difference to severity — a finding that complicates the widespread assumption that spaying and neutering uniformly reduce bite risk. Prior research has documented that intact male dogs bite more often, with one study reporting an incidence rate of 55.1 bites per 1,000 unneutered males versus 3.0 per 1,000 spayed females, and testosterone described as a behavior modulator that makes dogs react more quickly, more intensely, and for longer. But frequency and severity, this study suggests, are governed by different dynamics.

Perhaps the study’s most striking discovery concerns victim age. Rather than a simple linear decline in risk with each passing year, the relationship between age and bite severity turned out to be distinctly U-shaped. Each additional year of age was associated with a 5.6 percent decrease in the odds of a severe bite (odds ratio 0.944), while the quadratic term was also highly significant, indicating the risk curve bends back upward. The turning point sits at just over 47 years old: before that age, risk of severe injury falls as people get older; after it, risk climbs again. A sensitivity analysis excluding the oldest one percent of victims — those over 86 — confirmed the pattern was not an artifact of a few extreme cases. Even after controlling for bite location, the U-shaped age effect remained significant, suggesting that the very young and the elderly are intrinsically more vulnerable, not merely more likely to be bitten in dangerous places like the head and neck.

The authors frame these findings through a harm reduction lens, an approach borrowed from public health strategies that accept a behavior will persist and aim instead to blunt its worst consequences. Dog bites, they argue, will never be eliminated as long as humans keep companion animals, so the priority should be preventing the bites that cause the most damage. This matters legally as well as medically: under Colorado law, euthanasia becomes mandatory for dogs whose bites cause serious bodily injury or death, whereas France has adopted a more rehabilitative model requiring veterinary evaluation of biting dogs within 15 days to identify triggers and prevent recurrence. In Pueblo, a new ordinance creating a “potentially dangerous animal” category now gives enforcement officers flexibility to intervene before minor incidents escalate into tragic ones.

The practical implications are clear. Guardians, shelters, and adoption counselors should pay particular attention to households with young children and elderly members, the two groups at the highest risk of severe injury, and supervision of dogs around these vulnerable populations should be treated as a community responsibility rather than a burden placed on individual owners. The authors also recommend that policymakers consider making responsible pet ownership education mandatory at the point of adoption or purchase, rather than only after a court order following an incident. At the same time, the study acknowledges its limits: the administrative data lacked information on breed, dog size, victim sex, and race, and many bites — especially minor ones and those involving animal victims — go unreported, sometimes because people fear that reporting could doom a dog to euthanasia. Biting, the researchers emphasize, is a natural behavior with many motivations, from fear to territoriality to predatory instinct, and understanding the patterns behind the most severe outcomes is the first step toward a world where fewer bites turn tragic.

Subject of Research: Predictors of dog bite injury severity among human, dog, and cat victims in Southern Colorado

Article Title: Predictors of reported dog bite injury severity among human, dog, and cat victims in Southern Colorado

Article References: Briscoe, M. D., & Reszka, S. (2026). Predictors of reported dog bite injury severity among human, dog, and cat victims in Southern Colorado. Discover Animals, 3(1), Article 97. https://doi.org/10.1007/s44338-026-00256-1

Image Credits: AI Generated

DOI: 10.1007/s44338-026-00256-1

Keywords: dog bites, injury severity, public health, harm reduction, animal behavior, Colorado, logistic regression, victim age, neutering, animal law enforcement, veterinary science, risk factors