Machine learning predicts selected cat diseases using insurance data amid challenges in interpretability
Bibliographic record
Abstract
Objective: To develop models for prediction of the onset of specific diseases in cats using pet insurance data and to evaluate their predictive performance. Methods: Agria Pet Insurance data from almost 550,000 cats (2011 to 2016) were analyzed and used to train predictive models for periodontal disease and skin tumors using breed, sex, and insurance claim history. Random downsampling and 1:1 matching by age, insurance duration, and time at risk balanced the dataset. Variables were then further processed, with random forest and conditional logistic regression used for analysis. Model accuracy was assessed through leave-one-out cross-validation, while variable importance plots, partial dependence plots, and coefficients were used for model interpretation. Results: Model accuracy ranged from 81.9% to 88.2% (P < .01, baseline 50%). Key predictors included prior insurance claims for "digestive," "whole body symptom," "skin," and "injury conditions," which may be nonspecific and predictive of various diseases. Maine Coon, Siamese, and Burmese cats were associated with periodontal disease-positive predictions, while domestic cats were linked with negative predictions. For skin tumors, Norwegian Forest Cats, Devon Rex and Sphynx cats, and Maine Coon cats were associated with positive predictions, whereas Birman and domestic cats were linked with negative predictions. Conclusions: This study presents a method of machine learning predictive analysis on pet insurance data, although more comprehensive medical information and approaches accounting for data characteristics may be necessary to develop clearer predictors. Clinical Relevance: To prevent or detect these conditions early, veterinarians can use the breed risk results to guide clients, especially those with high-risk breeds, by offering early advice on lifestyle and monitoring.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".