The research team led by Prof. Victor O.K. Li and Prof. Jacqueline C.K. Lam from the Department of Electrical and Electronic Engineering (EEE), HKU, has won the US National Academy of Medicine (NAM) Healthy Longevity Catalyst Award (HLCA) 2024 for their project entitled “A Large Language Model (LLM)-driven Approach for Advancing Low-cost Timely Speech-based Late Onset Alzheimer’s Disease (LOAD) Prediction”. This is a joint award between Prof. Victor O.K. Li (PI, HKU-AI WiSe), Prof. Jacqueline C.K. Lam (Co-lead, HKU-AI WiSe), and Dr. Yang Han (AI, HKU-AI WiSe), Prof. Yam-leung Cheung (Linguistics, CUHK), Prof. James Rowe (Medical Science, Cambridge U), Dr. Jocelyn Downey (Biomedical Science, HKU), Prof. David Rubinzstein (Neuroscience, Cambridge U). This is the fourth time the team led by Victor and co-led by Jacqueline have won the 2024 US NAM HLCA award. The HLCA award is part of the Global Grand Challenge, a multiyear, multistage, and multimillion-dollar international competition designed to advance bold, novel ideas with the potential to dramatically improve health as people age.
Project Description:
Timely LOAD prediction is critical to slowing disease progression. Existing works based on clinical biomarkers are generally invasive, time-consuming and expensive, focussing on diagnosis instead of prognosis. In our multi-modal study, linguistic markers are strong early-stage predictors of LOAD. However, speech datasets are typically small, lacking longitudinal representation and linguistic variation, making it difficult to train LOAD models to accurately predict LOAD. Our study revolutionises LOAD prediction via the development of an LLM-driven transformer model trained on sufficient amounts of normal control (NC)/mild cognitive impairment (MCI)/LOAD speech samples. First, we develop a novel LLM-driven model to generate new speech and transcription samples based on limited NC/MCI/LOAD samples from DementiaBank and Framingham Heart Study. Particularly, more longitudinal Chinese and English speech samples of NC/MCI/LOAD will be generated from original samples to overcome data shortage. Second, our LOAD speech data are converted into speech and text embeddings capturing the linguistic information of pre-trained LLMs, subsequently fed into our Transformer for accurate prediction of timings for LOAD conversions (including pre-symptomatic LOAD), and identification of reliable linguistic markers discriminating these different conversion states. This has the advantage of increasing/enhancing (1) data volume, diversity, and temporal variability to improve accuracy, robustness and generalizability of LOAD prediction, (2) understanding of language-specific and cross-linguistic markers characterizing disease states (particularly Chinese and English contexts), (3) our detection of MCI/LOAD states, and prediction of the timings of pre-symptomatic onset. Our preliminary results show that LLM+Transformer (with-Generated-Transcription-Data) significantly outperforms Transformer Baseline (without-Generated-Transcription-Data) by 6% in NC/MCI/LOAD detection, achieving 91% accuracy.
Along with US$50,000 funding support for each award, the winners will gain exclusive access to additional funding opportunities and be entitled to join a global interdisciplinary network of innovators working towards achieving healthy longevity. International Catalyst Awardees and finalists who have made significant progress in advancing their ideas and preferably achieved proof of concept may apply for an Accelerator Award (US$185,000 to US$1 million, or more) and contend for the Grand Prizes (up to US$5 million) in subsequent phases of the Global Competition.
Congratulations to the winners of this esteemed accolade! The EEE Department wishes Victor and Jacqueline’s research team a great success in future funding competitions.
For more information, please visit the NAM website: https://healthylongevitychallenge.org/winners/a-large-language-model-llm-driven-approach-for-advancing-low-cost-timely-speech-based-late-onset-alzheimers-disease-load-prediction