How scientists use our data to learn about human health and improve patient experiences and outcomes

University of Birmingham academics and professionals from the health and community sectors invited members of the public to join them to discuss the potential of AI in healthcare. For example, analysing large volumes of health data can support treatment decisions, assist with hospital flow and planning, and enhance prescribing and diagnostic accuracy.

Sessions ranged from presentations to the whole group, to smaller activities and opportunities to ask questions and share ideas, including a creative opportunity with a local artist.

How AI supports healthcare

We can already see real world applications of AI in healthcare and the number of opportunities to use AI is only going to grow.

Researchers and members of the public were able to discuss how smartwatches collect data using sensors that can monitor heart rate and rhythm, skin temperature and other physiological signals to detect changes that might require further investigation. Another example includes smart patches worn by diabetics, which use AI to predict risks and customise treatment plans to individual patients in real time. Smart glasses are already in use in some settings to support training surgeons. These tools have the potential to revolutionise healthcare; saving resources, improving accuracy and making care more convenient for patients. However, the very definition of a ‘tool’ is that it helps the user complete a particular activity, it doesn’t do it for them. AI won’t be replacing doctors any time soon.

Optimising patient flow

Hospitals are getting busier, seeing more patients every year. To make best use of resources, minimise queuing and bottlenecks created by inefficient systems, AI can be employed to help optimise patient flow.

Visitors at the event watched simulation software set up hospitals in various ways to see how the number and placement of doctors and nurses could be varied to help get patients to the right place at the right time. Studies have already been using this software to test amendments before making recommendations to change real systems. It is hoped that patient benefits will soon be seen in real life.

Privacy considerations

Members of the public understandably have questions about who has access to their health data and how easily identifiable their information might be within a data set. There was plenty of opportunity for conversation on the topic of privacy and data security throughout the event.

Researchers shared how synthetic health data can be generated to scale up scarce patient data so that meaningful statistical conclusions can be drawn, whilst keeping patient information private. It also helps to provide researchers with data more quickly, without having to wait for permissions, meaning research can progress at pace.

Event participants also considered the significance of studying diverse populations for a better understanding of diseases and the importance of unconsented data access to ensure everyone is represented in the data, so that research helps to improve care for everyone.

Hands on activities

Members of the public were invited to get creative in a hands-on activity that sparked conversation about how we give information about our health to computers versus how we might share with a doctor and some of the challenges those discrepancies pose when analysing health data. For example, unstructured data recorded as part of a conversation with a doctor can be much harder to decode than tick boxes on a computer screen, however, advanced machine learning is starting to get to the bottom of how to categorise free flow fields in medical records.

The event rounded off with a panel discussion and an opportunity for audience members to put questions to the experts.

The panel, made up of Professor Dipak Kotecha, Professor Elizabeth Sapey, Suzy Gallier, Dr Xiaoxuan Lui and Professor Alastair Denniston, discussed the opportunities created by AI and health data for patients to take better control of their health, the risk that AI will further embed bias and perpetuate health inequalities, how using unconsented health data minimises this risk and helps ensure health advances benefit everyone, as well as the need for regulation to keep up with technological advances.