PROJECT CO-AUTHOR (MSC TEAM PAPER)

Medical LLM Reliability

Identified a 23.3 kappa-point multilingual reliability gap (English to Hindi) in 8B-parameter medical LLM specialists, across 6,000 total inferences.

Tech Stack
PythonAloe-Beta-8BMeditron3-7BGemini 2.5 Flash-LiteStatistical Analysis

Key Achievements

100 stratified USMLE-style questions x 10 runs x 2 languages = 6,000 inferences

Bootstrapped 95% confidence intervals with Bonferroni-corrected significance tests

Identified a 23.3 kappa-point EN-to-Hindi reliability gap in 8B specialists

Deep Dive

Accurate but Unreliable?

While Large Language Models (LLMs) are increasingly used in healthcare, most benchmarks focus solely on average accuracy. This co-authored MSc team study tackles a more critical, often ignored dimension: consistency.

If a clinical question is posed repeatedly, does the model give the same answer every time — and does that hold up across languages? For the roughly 600 million Hindi speakers worldwide, this “self-agreement” gap is a major barrier to safe telehealth adoption.


The Methodology: 6,000 Inferences, Two Languages

The team evaluated three medical LLMs — Aloe-Beta-8B, Meditron3-7B, and Gemini 2.5 Flash-Lite — on 100 stratified USMLE-style multiple-choice questions, each run 10 times in both English and Hindi: 6,000 total inferences.

Statistical Framework

Each of the 10 runs was treated as an independent rater, enabling:

  • Fleiss’ Kappa — measuring how consistently a model agrees with itself across runs.
  • Krippendorff’s Alpha — a chance-corrected reliability measure.
  • Bootstrapped 95% confidence intervals and Bonferroni-corrected Mann-Whitney U tests to confirm differences between languages were statistically significant, not noise.

The Result: A 23.3-Point Reliability Gap

The 8B-parameter specialist models showed a 23.3 kappa-point drop in self-agreement moving from English to Hindi — meaning the same clinical question, asked repeatedly in Hindi, was substantially more likely to get a different answer each time than the identical question asked in English. Average accuracy alone would never surface this: a model can look competent on a single pass while being meaningfully less trustworthy under repeated use.


Impact: Making AI Stable in Practice

This shifts the conversation from “is the AI smart?” to “is the AI stable?” — an essential distinction for deploying LLM-powered health tools to non-English-speaking populations, where a single wrong answer carries real clinical risk.