Translational Health Data Science and real time AI – bridging the gap between algorithms and healthcare application
Clinicians and data scientists at UCLH/UCL have designed and deployed the Experimental Medicine Application Platform (EMAP) to use artificial intelligence for patient benefit. Every week more than 1,000,000 data points feed algorithms to accelerate admissions from A&E, to improve antibiotic management and to better allocate ICU beds to patients in need.
EMAP is the product of co-investment by UCLH BRC and the UCLH Charity, and a collaboration between UCLH clinicians and UCL's Centre for Advanced Research Computing. It mirrors, in near real-time, the hospital's new Electronic Health Record System, employing a pioneering approach to protect patient privacy, developed in consultation with patients and the public. The system means information delivered to the patient’s bedside keeps pace with clinical and operational decision-making across the hospital.
Our algorithms are deployed to forecast emergency bed demand up to 24 hours in advance, allowing hospital bed managers to better prioritise who is discharged and allocate staff more efficiently. They can also identify patients who need specialist review of their antibiotics to reduce drug resistance and treatment failure (1).
EMAP is providing data for multiple clinical-academic teams funded by NHS-X. Our algorithms suggest that nine out of ten patients most likely to miss an appointment can be identified using cutting edge 'neural network' techniques applied to routine health data (2). The information is being used to target interventions that facilitate and encourage attendance at appointments.
The platform is also being used to optimise recruitment to clinical trials, for example in severe infection. This time-sensitive process is sped up by using a computer to scan ‘free text’ in thousands of clinical notes. The data is then cross-checked with blood tests and vital signs to assess patient suitability (3). This algorithm flagged almost ten times more patients than research nurse reviews, often many hours earlier. During the COVID pandemic, the EMAP tool has been used to help critical care teams deploy a real-time bed management tool.
Working with patients and the public to co-design a health data strategy, UCLH is at the forefront of using real time data science to bring benefits to patients and healthcare staff alike.
References:
1.Dutey-Magni et al JAC Antimicrob Resist 2021
2. Nelson A, NPJ Digital Medicine 2019
3. Tissot HC, IEEE J Biomed Health Inform 2020
