Artificial Intelligence (AI) for Healthcare and Planning: Scenario Modelling Future System Needs and Demands for HWE ICS

Background:
Pressures on health and care services are increasing. Demand continues to grow and understanding this and being able to quantify what needs to develop, and transform is essential to respond to this challenge and protect health and care services for those who are most vulnerable.
The Integrated Care System (ICS) in Hertfordshire and West Essex has agreed on a set of strategic priorities and outcomes to improve its population health, to be delivered via transformation programmes. Understanding forecasted demand across the system over the longer term and the current capacity to meet this, will help to inform our strategic, workforce and financial planning. The HWE population is expected to grow and age with a significant growth in the over 85s predicted.
Against this backdrop of challenges NHSE’s strategic direction is ICS’s will build an integrated care system intelligence function using data and analytics to enable effective decision-making.
“Data-driven population health approaches will be a key tool in our response to these inequalities. The challenge for ICSs will be to quickly develop their use of population health analytics from using data to inform their approach to condition management, to utilising predictive risk factors that help to increase early detection and prevent ill health and identify at-risk populations. Person and pathway-centred datasets, including information about the wider determinants of health, are needed, as well as the analytical teams and data-literate leaders that can make use of them. Data and analytics should not just drive smarter planning, care coordination, and performance management at the overall ICS level, but should also be supporting staff to deliver transformation on the front line.”

Benefits of partnership analytical work:
Advanced Analytics
• Models will identify system demand and capacity to support improved service delivery using advanced analytical techniques provided by skills from UH.
• Scenario modelling identifies ‘stretch’ from ICS transformation programmes to meet future demand whilst improving outcomes.
• Identification of further opportunities.
Workforce
• UH to deliver applied workplace training to upskill ICS data analysts in advanced analytical techniques.
• Short courses can be developed by UH to sustain development and support career advancement.

Expected Outcomes:
• Co-design and deliver demand and capacity predictive models for health and care services across HWE ICS targeting at risk populations whose outcomes can be improved in line with the HWE segmentation tool.
• For each care setting area consider a “do nothing” model, taking into account expected growth and changes in demographics and complexity.
• Through scenario modelling and stakeholder engagement, identify and model best practice planned changes to the way we plan and deliver care to develop forecasts that reflect a ‘realistic’ level of change to capacity.
• Feed in system outputs into the New Hospital Programme workstream.
• A skilled analytical workforce able to meet the future system analytical demands as described within ‘What Good Looks Like’

UH Solution
• The School of Physics, Engineering, and Computer Science (SPECS) will develop artificial intelligence (AI) tools for the support of the data management and planning needs of the HWEICS.
• The AI tools will forecast demand across the system using predicting the demand and capacity needs of the healthcare system.
• UH/SPECS team will develop AI models for the modelling of patients behaviour, grouping of them and forecasting of patients admission risk or emergency attendance.
• For the development of the AI models, we will rely on state-of-the-art technology for time series analysis and forecasting using machine learning algorithms like SVM, Decision Trees, as well as deep learning algorithms like LSTM and RNN. Data augmentation using synthetic data generative AI models will be considered as well.
• The AI predictive models will consider patient population groups and their time dynamics, especially the predicted population growth of the over 85’s. The developed AI tools will meet the NHSE requirement for a system intelligence function and be integrated into a framework environment to facilitate effective decision-making.
• Taking an innovative approach to supporting the ICS in retaining its analytical workforce.
• SPECS propose to develop a novel applied training programme. Using data projects from the ICS transformation programmes, an embedded post-doctoral expert analyst will create a bespoke training programme for ICS workforce working within predictive analytics and data science. It is proposed that the ICB host the UH analyst so ICS staff can be trained on their site, with their data, using their systems on ICS projects. This creates sustainability and relevance. The 2021 analytical skills audit will be used to ensure the identified development gaps are in the training plan.