Medical Decision Support

Medical Decision Support is a project funded by Precision Health FSP and Data61. It covers a range of activities

  • Clinical trials search: Recruitment for clinical trials is non-trivial. Clinical trials which for example examine the efficacy of a treatment for a particular disease, require patients that match specific inclusion and exclusion criteria. We develop a system to assist clinicians and patients in finding clinical trials that match. Our system deploys a combination of Natural Language Processing (NLP), Machine Learning (ML), and Information Retrieval (IR) techniques to find the most suitable ranking and scoring methods for this complex problem.
  • Biomedical literature search: Clinical decision support using the latest evidence published in the biomedical literature. We developed A2A benchmarking system to allow NLP/IR researchers working on biomedical search to compare the most common query formulation and text processing techniques on several biomedical datasets released by NIST/TREC over the years.
  • Patient-focused question answering: Patients, similar to physicians, require answers to their biomedical questions. We develop techniques to allow patients to find answers using a combination of NLP, IR, and deep learning techniques.

  • Information extraction for proteomics: This is a collaboration with Flemish Institute of Bietechnology (VIB) and the Polish Academy of Sciences (PAN). Our goal is the identification of Post Translational Modification (PTM) information from scientific text. We use information extraction techniques to identify protein, gene, residue and position mentions in scientific texts.

International challenges (shared tasks)

We participated in the following challenges over the years:

Publications:

Reports: 

Presentations:

  • CSIROmed at TREC Precision Medicine at TREC 2020 conference, November 2020 (Speaker: Maciej Rybinski)
  • From Information Retrieval Benchmarking to Precision Medicine Knowledge Discovery, Joint NETTAB/BBCC 2019 Meeting (Speaker: Maciej Rybinski)
  • AI in Health, CSIRO Connect 2019 (Speaker: Sarvnaz Karimi)

Prototypes and Resources:

  • A2A (decision support): a2a.csiro.au
  • A2A (literature search): a2a.csiro.au/benchmarking/
  • S2C (Clinical trials): a2a.csiro.au/s2c
  • A video explaining what Science2Cure (S2C) search prototype does!
  • Medical research article summarisation and evaluation (domain experts to evaluate the quality of LLM-generated summaries, focusing on the faithfulness dimension)
  • MedRedQA Dataset
A2A System

A2A diagram (from BMC Bioinformatics)

 

Team: 

Medical Decision Support Team

Medical Decision Support Team, 2020-2021

Team photo - May 2022

Precision Health FSP Project team, 2021-2022

We also collaborate and benefit from the advice and input of our colleagues at CSIRO and our university partners, including: Cecile Paris (CSIRO Data61), Stephen Wan (CSIRO Data61), Nathan O’Callaghan (PH FSP Leader, CSIRO Health & Biosecurity), Brian Jin (CSIRO Data61), Anthony Nguyen (CSIRO Health & Biosecurity), Sara Falamaki, Zhenchang Xing (ANU).

Past Students:

  • Jerry Xu (Summer Intern, Sydney University, Nov 2019–Feb 2020)
  • Aleney Khoo (Summer Intern, Sydney University, Dec 2020–Feb 2021)
  • Jeffrey Yang (Summer Intern, UNSW, Dec 2020–Feb 2021)
  • Sonit Singh (PhD student, CSIRO Data61 and Macquarie University, graduated 2021)
  • Liam Watts (Summer intern, Nov 2021–Jun 2022)
  • Mong Yuan Sim (Summer intern, Adelaide University, Nov 2022–Mar 2023)
  • Anqi Wu (Summer intern, Monash University, Nov 2022–Feb 2023)
  • Skyler Zou (Summer intern, UNSW, Nov 2022–Mar 2023)