2024

• (19 December 2024) Paper published on npj Digital Medicine - Optimising the paradigms of human AI collaborative clinical coding - at DOI:10.1038/s41746-024-01363-7
  
  Clinical coding is the process of assigning standardised codes (eg hashtag#ICD-10 for diagnosis or procedures) for an interaction with the health service (a visit to GP or a hospital stay). Such ‘coded’ information is widely used for patient care, auditing and research. However, currently this is mainly a manual process done by clinical coders, which is expensive, time-consuming, and error-prone (quantified in the paper). Yue Gao, a visiting PhD student at KnowLab during 2023-2024, developed a deep-learning based human-AI collaborative paradigm to facilitate the clinical coding process. The system, called CliniCoCo, was developed using real-world hashtag#EHR datasets from two hospitals in China and further evaluated by human coders in a third Chinese hospital. CliniCoCo was able to reduce coding time by 40% and enable professional coders achieving a F1 score more than 0.93.
  
  Read it at DOI:10.1038/s41746-024-01363-7.

• (28 October 2024) Invited talk at KnowLab - The world of non-clinical safety, artificial intelligence, and foundation models
  
  Speaker - Dr Arijit Patra
  
  Title - The world of non-clinical safety, artificial intelligence, and foundation models
  
  Abstract - There has been a significant amount of interest in artificial intelligence in the pharmaceutical industry, particularly with respect to drug discovery. While drug discovery has been established as a domain where algorithmic involvement can create substantial efficiency gains, other aspects of the pharmaceutical development process such as non-clinical and clinical safety are of critical importance towards creating patient value. This talk will explore the drug development process, the concepts of non-clinical safety with a particular focus on toxicologic pathology, and recent insights on building ML pipelines and foundation models therein.
  
  Bio - Dr Arijit Patra is a Senior Principal Scientist at UCB Biopharma UK. He holds a PhD in machine learning for healthcare imaging from the University of Oxford, where he was a Rhodes Scholar (India & Exeter, 2016). Prior to that, he completed a dual degree in Mechanical Engineering from the Indian Institute of Technology (IIT) at Kharagpur, India. He has also been associated with AstraZeneca, Shell, Microsoft Research and CSIR-South Africa at various points in his career and has been actively involved in the AI4SG (AI for Social Good) community. He has authored several publications around machine learning and medical imaging and is a reviewer for multiple peer reviewed venues such as NeurIPS, ICML, MICCAI and several journals. Arijit has been appointed as a Rising Leaders fellow of the Aspen Institute UK, and as an International Strategy Forum fellow in 2024.
  
  Date/Time - 13.30-14.30 14th November 2024

• (10 October 2024) Tutorial on Foundation Models For Medical Imaging - with MICCAI 2024, Morocco
  
  Generative AI and large-scale self-supervised foundation models are poised to have a profound impact on human decision making across occupations. Healthcare is one such area where such models have the capacity to impact patients, clinicians, and other care providers.
  
  As part of MICCAI2024, Yunsoo, Jinge and Honghan worked with colleagues from IBM, ÉTS Montréal, and Stanford University to organize a tutorial on Foundation Models For Medical Imaging. Yunsoo and the rest of KnowLab team took the lead on “the multimodal LLMs in medicine” section.
  
  Tutorial website at https://sites.google.com/view/miccai-2024-tutorial/home.

• (8 October 2024) Paper published on BMC Medical Informatics and Decision Making - A hybrid framework with large language models for rare disease phenotyping - at DOI:10.1186/s12911-024-02698-7
  
  Rare diseases pose significant challenges in diagnosis and treatment due to their low prevalence and heterogeneous clinical presentations. Unstructured clinical notes contain valuable information for identifying rare diseases, but manual curation is time-consuming and prone to subjectivity. This study aims to develop a hybrid approach combining dictionary-based natural language processing (NLP) tools with large language models (LLMs) to improve rare disease identification from unstructured clinical reports. The proposed hybrid approach demonstrates superior performance compared to traditional NLP systems and standalone LLMs. LLaMA3 and Phi3-mini achieve the highest F1 scores in rare disease identification. Few-shot prompting with 1-3 examples yields the best results, while knowledge-augmented generation shows limited improvement. Notably, the approach uncovers a significant number of potential rare disease cases not documented in structured diagnostic records, highlighting its ability to identify previously unrecognized patients.
  
  Read it at DOI:10.1186/s12911-024-02698-7.

• (3 August 2024) Paper accepted by the International Workshop on Trustworthy Artificial Intelligence for Healthcare 2024 - Human-in-the-Loop Chest X-Ray Diagnosis - Enhancing Large Multimodal Models with Eye Fixation Inputs - at DOI:10.1007/978-3-031-67751-9_6
  
  In the realm of artificial intelligence-assisted diagnostics, recent advances in foundational models have shown great promise, particularly in medical image computing. However, the current scope of human-computer interaction with these models is often limited to inputting images and text prompts. In this study, we propose a novel human-in-the-loop approach for chest X-ray diagnosis with a large language and vision assistant using eye fixation prompts. The eye fixation prompts contain the location and duration of a radiologist’s attention during chest X-ray analysis. This assistant interacts with a radiologist in two ways - diagnosis recommendations of possible diseases and diagnosis report confirmation. The results show the enhanced human-computer interaction with the eye fixation prompt significantly improves the accuracy of the large multimodal model’s performance in differential diagnosis and report confirmation. Fine-tuning with just 658 reports with fixation information further boosted the performance of the LLaVA-1.5, surpassing the previous state-of-the-art model LLaVA-ERR, which was trained on 17k MIMIC reports, by 5%. Our study highlights that this novel approach can better assist radiologists in clinical decision-making in a reciprocal interaction where the models also benefit from the domain expertise of radiologists.
  
  Read it at DOI:10.1007/978-3-031-67751-9_6.