arXiv subjects

Anthonio Oladimeji Gabriel

Publications and source records attributed to Anthonio Oladimeji Gabriel.

4 recordsLinked to original sources

arXiv2026

IyawoBench v2.0: Extended Diagnostic Evaluation of Large Language Model Clinical Triage in Nigerian Primary Care

Large language models are being deployed as clinical triage tools in low and middle income countries where trained physicians are scarce. Existing safety metrics, however, produce misleading confidence: models scoring 100% on binary "did not send an emergency home" safety measures may nevertheless exhibit systematic failure modes that render them undeployable at scale. We present IyawoBench v2.0, an extended diagnostic evaluation of large language model clinical triage on 200 synthetic vignettes derived from 1,200 real patient encounters at 19 Nigerian primary health centres. We introduce a formal mathematical framework comprising fourteen definitions and two theorems that decompose triage safety into three distinct failure modes: Conservative Escalation Bias, Systematic Downgrade Bias, and Middle-Tier Instability. We propose the Escalation Bias Index and Expected Deployment Cost as novel metrics that expose failure modes hidden by conventional accuracy and sensitivity scores. Evaluated on three frontier models (Claude Sonnet 4.6, Llama 3.3 70B, Llama 3.1 8B) plus five naive baselines, we show that: (1) all three models exhibit at least one formal failure mode; (2) traditional sensitivity metrics conceal a 77 percentage point under-triage gap in Llama 3.1 8B; (3) the optimal model varies across three deployment scenarios (Emergency-Focused, System-Sustainability, Balanced), demonstrating that single-ranking benchmarks are inadequate for LMIC clinical AI selection. IyawoBench v2.0 provides both a rigorous benchmark and a diagnostic framework transferable to any triage-style clinical AI evaluation. All code, data, and analysis pipelines are publicly available.

cs.CY↗

arXiv2026

Safety That Does Not Transfer: Cross-Lingual Clinical Correctness Drift in Deployable Medical Language Models

Safety evaluation of large language models is conducted predominantly in English and predominantly on frontier systems. Neither condition describes how such models are encountered in low-resource health settings, where small quantised systems are run locally and queried in local languages. We ask whether clinical safety established in English transfers to Hausa, and whether any failure is attributable to the language, the clinical task, or the class of model that low-resource deployment admits. Matched English-Hausa question pairs were built for three conditions of high burden in northern Nigeria: malaria, sickle cell disease, and tuberculosis, probing knowledge recall, emergency triage, a leading question inviting a contraindicated action, and a traditional-remedy claim. Six models were evaluated: five locally deployable systems of 4-9 billion parameters, two medically fine-tuned, and one frontier system. All 128 responses were scored against Nigerian national treatment guidelines by two fluent Hausa speakers working independently and blind to one another. Among locally deployable models, mean clinical correctness fell from 1.57 in English to -0.03 in Hausa, on a scale where 2 denotes a correct answer and -1 an actively harmful one. The frontier model moved from 2.00 to 1.75 and produced no response judged harmful in either language. Drift was consistent across all three conditions. Inter-rater agreement was substantial for clinical correctness (kappa = 0.70); agreement on harm was initially poor (kappa = 0.22) and is examined in detail. Because a frontier model answers the same questions competently in Hausa, the deficit is a property neither of the language nor of the clinical material, but of the deployable tier.

cs.CL↗

arXiv2026

IyàwóBench: A Benchmark for Evaluating Large Language Model Clinical Triage Accuracy on Undifferentiated Febrile Illness in Nigerian Primary Health Settings

Background. Undifferentiated febrile illness is the leading cause of primary care outpatient visits in Nigeria, yet no validated benchmark exists for evaluating large language model (LLM) clinical triage reasoning in West African primary health settings. Methods. We introduce IyàwóBench v1.0, a dataset of 200 synthetic clinical vignettes across eight febrile illness categories derived from statistical distributions of 1,200 real patient encounters at 19 primary health centres (PHCs) in Oyo State, Nigeria. Six LLMs were evaluated on structured triage classification across two metrics: triage accuracy and safety score. Results. All six models achieved 100% safety scores (95% CI: 96.4-100.0%), never downgrading a critical REFER NOW case to TREAT HERE. Triage accuracy varied substantially: Claude Sonnet (claude-sonnet-4-5) 67.5% (95% CI: 60.8-73.7%), Llama 4 Scout 59.5% (52.5-66.2%), Llama 3.3 70B 43.0% (36.2-50.0%), and Llama 3.1 8B 39.0% (32.4-45.9%). Two models demonstrated near-zero accuracy attributable to structured output non-compliance. Conclusions. Modern LLMs exhibit safe triage behaviour but vary substantially in structured clinical accuracy. Clinically engineered systems with embedded WHO guidelines outperform general-purpose models by up to 28.5 percentage points. IyàwóBench provides the first reproducible evaluation framework for LLM clinical decision support in West African primary care.

cs.CY↗

arXiv2026

Adversarial Fragility and Language Vulnerability in Clinical AI: A Systematic Audit of Diagnostic Collapse Under Imperceptible Perturbations and Cross-Lingual Drift in Low-Resource Healthcare Settings

Current clinical artificial intelligence (AI) systems are evaluated almost exclusively on clean, standardised, English-language inputs, conditions that do not reflect the realities of healthcare delivery in low-resource settings. This study presents the first systematic dual audit of two orthogonal safety vulnerabilities in clinical AI: adversarial image fragility and cross-lingual diagnostic drift. Using DenseNet121, the architecture underlying CheXNet, fine-tuned on the COVID-QU-Ex chest X-ray dataset (85,318 images; COVID-19, Non-COVID Pneumonia, Normal), we demonstrate that diagnostic accuracy collapses from 89.3% to 62.0% under a Fast Gradient Method (FGM) perturbation of epsilon=0.021, a magnitude imperceptible to the human eye. Standard defensive strategies including Gaussian smoothing and ensemble voting failed to restore clinical safety. In a parallel language fragility experiment, we tested Llama3.1:8b and NatLAS (N-ATLAS) on 20 COVID-19 clinical cases presented in Standard English, Nigerian Pidgin (Naija), and Yoruba-inflected English. Both models exhibited significant accuracy degradation: Llama3.1:8b dropped from 80.0% to 65.0% on Pidgin; NatLAS, an African-context model, collapsed from 85.0% to 55.0%, with diagnosis consistency falling to 50%. These findings establish a quantitative failure envelope for clinical AI under conditions representative of Primary Health Centre (PHC) deployment in Nigeria, and motivate urgent calls for adversarially hardened, linguistically inclusive clinical AI architectures.

cs.CY↗