analiza aiqbench
This study applies interpretable machine learning techniques, specifically Naive Bayes classifiers, to evaluate the predictive utility of inflammatory biomarkers for cognitive impairment in an older Hispanic cohort. The research highlights a critical methodological distinction between statistical significance and actual predictive value, demonstrating that I-309 (CCL1) provides robust incremental prediction while maintaining model transparency. These findings offer valuable insights into best practices for handling small clinical tabular datasets where deep learning models are often impractical or opaque.
Machine-Learning Assessment of the Predictive Value of Inflammatory Biomarkers for Cognitive Impairment in an Older Hispanic Adult Cohort
arXiv:2609.19374v1 Announce Type: new Abstract: Small clinical tabular datasets require interpretable machine learning because deep learning is often impractical and ensemble models can be difficult to inspect. A key pitfall is that statistical significance does not necessarily imply predictive utility. Using data from the Panama Aging Research Initiative--Health Disparities (PARI-HD) cohort (n=165), we implemented a leakage-safe threshold-likelihood Bernoulli/Categorical Naive Bayes (BNB/CNB) classifier. Within every training fold, each continuous predictor was reduced to a supervised chi-square-derived state, while income entered the model through a categorical likelihood. All data-dependent steps were performed within repeated stratified 10-fold cross-validation with 30 repeats. The demographic baseline achieved a ROC-AUC of 0.630 +/- 0.017. I-309 (CCL1) was the dominant incremental feature, increasing AUC by 0.110, with paired DeLong tests yielding p<0.05 in 100% of repeats. In the pre-specified primary analysis, I-309 produced a fixed-partition DeLong p=0.0018, with robustness assessed across 200 random partitions, where the median p-value was 0.0011. Within the exploratory family of 18 candidate markers, I-309 achieved a Benjamini-Hochberg-adjusted q=0.032 on the frozen partition and satisfied q<0.05 in 85% of random partitions, whereas no other marker demonstrated reliable incremental predictive value. Because the fitted model is an inspectable table of thresholds and class-conditional probabilities, these results identify I-309/CCL1 as an interpretable candidate feature for tabular prediction of cognitive impairment, pending external validation.
Abstract:Small clinical tabular datasets require interpretable machine learning because deep learning is often impractical and ensemble models can be difficult to inspect. A key pitfall is that statistical significance does not necessarily imply predictive utility. Using data from the Panama Aging Research Initiative--Health Disparities (PARI-HD) cohort (n=165), we implemented a leakage-safe threshold-likelihood Bernoulli/Categorical Naive Bayes (BNB/CNB) classifier. Within every training fold, each continuous predictor was reduced to a supervised chi-square-derived state, while income entered the model through a categorical likelihood. All data-dependent steps were performed within repeated stratified 10-fold cross-validation with 30 repeats. The demographic baseline achieved a ROC-AUC of 0.630 +/- 0.017. I-309 (CCL1) was the dominant incremental feature, increasing AUC by 0.110, with paired DeLong tests yielding p<0.05 in 100% of repeats. In the pre-specified primary analysis, I-309 produced a fixed-partition DeLong p=0.0018, with robustness assessed across 200 random partitions, where the median p-value was 0.0011. Within the exploratory family of 18 candidate markers, I-309 achieved a Benjamini-Hochberg-adjusted q=0.032 on the frozen partition and satisfied q<0.05 in 85% of random partitions, whereas no other marker demonstrated reliable incremental predictive value. Because the fitted model is an inspectable table of thresholds and class-conditional probabilities, these results identify I-309/CCL1 as an interpretable candidate feature for tabular prediction of cognitive impairment, pending external validation.
Machine Learning (cs.LG)
arXiv:2609.19374 [cs.LG]
(or arXiv:2609.19374v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.19374 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 307 cuvinte
Subiecte conexe
Automating Learner Assessment: Benchmarking Machine Learning and Deep Learning Models for EEG-Based Familiarity Prediction
arXiv · 19 aug. 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · 18 sept. 2026
Explainable Graph-theoretical Machine Learning with Application to Alzheimer's Disease Prediction
arXiv · 16 sept. 2026
Early Detection of Alzheimer's Disease Using Explainable Machine Learning on Clinical Biomarkers: A Multi-Class Classification Study Using the Alzheimer's Disease Neuroimaging Initiative (ADNI) Dataset
arXiv · 5 iun. 2026