Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: DRA : Canada Graduate Research Scholarship - Doctoral Program (CGRS D)

McGill University

$120,000.00

Uncertainty Aware Large Language Models for Clinical Decision Support in Early Diagnosis of Oral Cancer.

EF 2025-26

Résumé
Oral cancer is becoming more common in North America, but it is often missed during routine dental visits. Early detection is critical because it greatly improves survival, yet many oral lesions are overlooked due to gaps in training and outdated knowledge among clinicians. Dentists already record a detailed information in patient clinical charts, but this information is rarely analysed in ways that could help detect cancer earlier. New artificial intelligence (AI) tools called large language models (LLMs) can read and understand written text, and they have shown great results in solving medical problems like interpreting X-rays. However, these models can sometimes give wrong answers with high confidence, which makes them unsafe to use in healthcare without additional safeguards. This project will create the first AI model designed specifically in field of dentistry to help clinicians identify early signs of oral cancer from patient records. We will: 1.Build a new, high-quality dataset from dental textbooks, scientific papers, and patient notes carefully reviewed by expert clinicians to train the AI model. 2.Train and fine-tune AI model on this dataset so it can recognise patterns linked to oral cancer more accurately. 3.Develop new methods to measure and communicate the AI model's uncertainty, so clinicians know when to trust its suggestions and when to be cautious. The outcome of this project will be an open-source AI tool that combines accuracy with reliability, giving clinicians better support in diagnosing oral cancer early. This could lead to earlier treatment, improved survival, and reduced the economic burden of disease for patients and families and healthcare system.
Programme
DRA : Canada Graduate Research Scholarship - Doctoral Program (CGRS D)
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
CIHR Funding Decisions Database

Tel que publié

Id
560187
Name
Sharma Mridul
Region
Québec
Country
Canada
Orgtype
Institution Paid; Research Institution
Instname2
Musculoskeletal Health and Arthritis
Theme2
Health systems / services
Prcname2
DT2 (Health services and population health research) CGRS Doctoral
Projecttitle
Uncertainty Aware Large Language Models for Clinical Decision Support in Early Diagnosis of Oral Cancer.
Programtype2
Studentship Programs
Orgnameinp2
McGill University
Cihramount2
$120,000
Cihrequipment2
$0
Programname2
DRA : Canada Graduate Research Scholarship - Doctoral Program (CGRS D)
Keyworddelim
Approximate Bayesian Computation; Clinical Decision Support; Early Cancer Detection; Fine Tuning; Large Language Models; Oral Cancer; Uncertainty Quantification
Abstract
Oral cancer is becoming more common in North America, but it is often missed during routine dental visits. Early detection is critical because it greatly improves survival, yet many oral lesions are overlooked due to gaps in training and outdated knowledge among clinicians. Dentists already record a detailed information in patient clinical charts, but this information is rarely analysed in ways that could help detect cancer earlier. New artificial intelligence (AI) tools called large language models (LLMs) can read and understand written text, and they have shown great results in solving medical problems like interpreting X-rays. However, these models can sometimes give wrong answers with high confidence, which makes them unsafe to use in healthcare without additional safeguards. This project will create the first AI model designed specifically in field of dentistry to help clinicians identify early signs of oral cancer from patient records. We will: 1.Build a new, high-quality dataset from dental textbooks, scientific papers, and patient notes carefully reviewed by expert clinicians to train the AI model. 2.Train and fine-tune AI model on this dataset so it can recognise patterns linked to oral cancer more accurately. 3.Develop new methods to measure and communicate the AI model's uncertainty, so clinicians know when to trust its suggestions and when to be cautious. The outcome of this project will be an open-source AI tool that combines accuracy with reliability, giving clinicians better support in diagnosing oral cancer early. This could lead to earlier treatment, improved survival, and reduced the economic burden of disease for patients and families and healthcare system.
Competitiondate
202511
Approvedterm2
3 yrs 0 mth
Pinamesdelim
Sharma, Mridul
Approvedterm
3 ans 0 mois
Cihrcontribution2
~ $120,000~ $0~