Paris hospitals use big data to predict admissions

The Assistance Publique-Hôpitaux de Paris group of hospitals is trialling a machine learning analytics system to better manage high demand of health resources.

Ten years of Paris hospital admissions data, plus external information including weather, patterns of flu infection and public holidays, was ingested and analyzed by the Trusted Analytics Platform (TAP). Four of the Assistance Publique-Hôpitaux de Paris (AP-HP) are trialling the new predictive system aimed at reducing hospital waiting times. TAP is an open-source analysis platform that uses machine learning to create tools and services across a range of industries.

The AP-HP project is the first to use open source time series analysis, and the TAP engineers that worked on the system had to build new sets of code. If the project is successful, plans are to implement the prediction system through the entirety of the 44 hospitals in the AP-HP group. The AP-HP system is accessed via a web browser and is being used by clinical and administrative staff, allowing admission rates to be forecast up to 15 days in advance. Ideally, this will provide enough advance warning for extra staff to be made available when demand is likely to be high.

Healthcare is embracing technology, looking for new ways to be more efficient with limited resources. A new AI-powered medical information system provides more accurate information for people at home, and an app is being used in the UK to provide clinicians with real-time, customizable visualizations of data. What areas of health are still awaiting a technology update?