PARIEDA - Prediction of Acute Respiratory Infection outcomes prior to Emergency Department Attendance
Principal Investigator: Dr Daniel Burns, Senior Research Engineer, University of Southampton, firstname.lastname@example.org
Professor Michael Boniface, Professorial Fellow of Information Systems, University of Southampton, email@example.com
Professor Matthew Inada-Kim, National Clinical Director-Infection, AMR & Deterioration-NHS England & Improvement, National Clinical Lead COVID NHS@home Visiting Professor, University of Southampton Chair COVID pathways group, firstname.lastname@example.org
Dr Stephen Kidd, Lead Healthcare Scientist, Hampshire Hospitals Foundation Trust,email@example.com
Aim: We will help community doctors and nurses decide how best to care for patients with serious respiratory illness. The right care depends on how ill a patient is and if they will get worse. Care may include home monitoring or hospital visits. We aim to use computer algorithms to help doctors and nurses make these decisions. We expect patients to avoid unnecessary trips to hospital and to feel more supported.
Background: Hospitals have had record number of emergency departments visits. Respiratory infections are almost half of the visits. Many of these patients were not admitted to hospital. This means that some patients could be cared outside of the hospital in the community. COVID-19 is a serious respiratory illness. During the COVID-19 pandemic a new way to care for patients was created. Instead of patients going straight to hospital, they were assessed in the community. Only the most serious cases were then sent to hospital. Community care and assessment is now being considered for other respiratory illnesses.
Approach: We will use computer algorithms to help community doctors and nurses decide which patients are at most risk of serious respiratory illness. Risk assessment will be done using machine learning. Machine learning is a way to train a computer to categorise patients into groups using data about patients and services they use. We will use historical hospital data to identify patients in high-risk groups. The patient categories will then be used to inform community decisions before attendance at hospital.
Patient and Public Involvement: Patients and public have helped develop the research through evaluation pilots for community assessment hubs. PPI will influence data usage and the use of risk groupings within care pathways. Two public members will participant in a Steering Committee. A PPI Committee will organise three workshops involving ten patients and public in the research.
Dissemination: Communication will engage the public and decision makers. We will work with patients and the public to design engaging communication and seek acceptance. Our results will be published and will inform national policy.