The 2024 Australian Medivate Data Challenge has been held in Sydney and Perth, with 35 students connecting with industry professionals, and leaders in the medical field, to develop innovative and novel solutions real-world health data challenges.
The Australian Medivate Data Challenge is the result of collaboration between The University of Sydney Society of Medical Innovation (SUSMI), the Digital Health CRC, and global AI healthtech, Beamtree. This year, the event was also held in Perth, hosted by the University of Western Australia’s Biomedical Engineering Society and Data Science Club.
Held from 1-4 July, the competition saw nine teams (five in Sydney and four in Perth) apply their creativity and technical skills to solve a contemporary healthcare challenge, culminating in pitching their cases before a panel of judges.
This year’s theme was Unlocking Health Data, Analytics and AI, with students tasked with creating and pitching a digital health product from one of three broad themes: Clinical Decision Support, Resource Optimisation or Consumer and Equity.
The hackathon is designed to bring STEM students with diverse backgrounds together, providing an opportunity to collaborate, learn and innovate, both with other students as well as industry professionals. This year, students came from a range of disciplines including biomedical engineering, data science, neuroscience, medicine, psychology, statistics and computer science.
The National Judging Panel, consisting of Culture Pilot Co Founder, Teresa Lily, Head of Analytics and Insights at Beamtree, Lachlan Rudd, and harrison.ai Clinical AI Engineer, Chris Chiu, assessed pitches from the four finalists (two teams from Sydney and two from Perth) awarding the title of 2024 Australian Medivate Champions to University of Sydney students Kevin Hou and Matthew Shu.
The winners received their due credit as well as a first prize of $700.
The 2024 Medivate Finalists
The winners, Kevin Hou and Matthew Shu conceived Nightingale AI, a healthtech that would build an analytics dashboard to enable Australian healthcare systems to evaluate AI/ML models – with the key goals of mitigating bias, maintaining fairness, benchmarking clinical utility, and evaluating performance. The platform would serve regulators thinking about safe AI deployment, hospitals adopting effective technology, and researchers/developers bridging the research-to-deployment chasm.
Runner’s Up, Glucode, comprised Alaleh Saeidi, Jiarui Zhang, Alqudus Lawal, Ashvini Pavalahandran. They identified the challenge and stress facing children with type 1 diabeties in planning a diet which keeps glucose levels under control. Recognising Continuous Glucose Monitors aren’t being fully utilised for their real-time diet feedback potential, the Glucode app combines AI, computer vision, barcode scanning and machine learning to analyse how diet impacts glucose levels. By tapping into CGM data, users would unlock personalised meal suggestions – partnering with grocery and food delivery to bring vouchers and convenience, as well as a gamified points system to reward users for healthy choices. With timely alerts, the ability to scan and log meals at a glance, and personalised meal schedules, recipes and checklists, Glucode helps “Balance Sweets for Sweeter Smiles”.
Perth-based team Aimee Soudre, Timothy Lau, Vanessa Miller presented a model that would help reduce hospital readmissions. The algorithm envisaged would work alongside the medical team in the hospitals during the discharge process. As the patient stays in the hospital, relevant data (especially time-based data) that is indicative of a patient’s health status would be fed into an algorithm (known as LSTM + CNN model) which will then determine if the patient is or isn’t eligible for discharge. The eligibility for discharge, risk of readmission percentage and abnormalities for the patient is displayed on a separate interface that the doctor would have access to and could use to help better inform the decision to discharge or not.
The second team in the finals from Perth were Jayashini M Patmanathan, Jia Yu Lau, Chloe Santos, Yanxi Liu who worked on an idea aimed at improving automated clinical coding systems. The solution looked to address the inefficiency and the inaccuracy of documentation by digitalizing documentation through a software interface using designated tablets. The software would use Natural Language Processing (NLP) and Machine Learning to predict text, prompt for specific information, flag incomplete documentation, and prompt further clarification. Once medical personnel have completed the document, the system generates a summary of the inputted information for further approval by medical personnel to ensure accuracy. Based on the feasibility study conducted, it was found that, assuming a project lifespan of 12 years, the rate of return on investment was 2.2 years with the hospital saving $2.9 million dollars from billing errors and labour costs.
