A head injury in an older person is now more likely to come from a standing-height fall than from a car crash, yet the tools doctors use to predict recovery were built on data from younger patients with high-impact injuries. This matters because neurosurgeons may be overly pessimistic about outcomes for older patients, potentially withholding surgery or intensive care based on age alone. Current prognostic calculators ignore frailty and multiple long-term health conditions, and they rely on subjective assessments of brain scans. The research team has already built a pipeline that links routine NHS data with automated analysis of brain imaging for other conditions, and they plan to apply the same approach to head injury. If successful, this could produce a more accurate, data-driven calculator that accounts for an individual’s baseline health, frailty, and scan results. That would help clinicians and families make better-informed decisions about whether transfer for surgery or intensive care is worthwhile, rather than relying on age as a proxy. The project is not fundamental science—it is a direct attempt to improve a clinical prediction tool that currently fails the patients it most often sees.
View original technical description
The demographics of head injury patients has changed remarkably in the last few decades, with standing height falls in older people now the most common cause (1). As our population ages, recovery from head injury is as much about background multi-morbidity as it is about the injury itself and age, co-morbidities and frailty contribute significantly to outcome (1, 2, 3, 4). However, how clinicians use these factors to drive decision making is less clear, and it has been recognised that neurosurgeons can be overly nihilistic in predicting outcome from head injury (5). Only 5% of patients over 60 have a neurosurgical intervention following a head injury, compared to 13% of younger patients (1). Whilst there are multiple factors in surgical decision making, it is important that interventions which improve outcome are not withheld based on simple judgements of age alone. Clinicians are always balancing risk and benefit when considering whether to transfer for surgery/intensive care, guided by previous trial data on outcomes. However, trials in the widely used IMPACT predictive calculator for head injury almost exclusively excluded patients over 65 or 70 (6) and was validated on the CRASH data-set of primarily younger patients with high-impact injury (7).This makes the data less applicable to the older population and there is no consideration of frailty or multi-morbidity (8). Current calculators are out-dated and incorporate subjective user-assessment of brain imaging (e.g. estimates of pattern/volume of blood and brain shift). There is a research gap in understanding the overall impact of age-related changes, frailty and health and social problems on outcomes in head injury patients. This could be answered by harnessing the power of data integration and machine learning to train better prognostic calculators on clinical and radiological data. Our research group (SustAIn SW) has a track record of using artificial intelligence (AI) to integrate routine care data and imaging in stroke prediction, Cauda Equina Syndrome and Parkinson s Disease (10-13). We have established a pipeline for clinical imaging transfer across Southwest Peninsula to the University with auto-anonymisation so that it can be digitally analysed. We have acquired over 100,000 patient scans and 60,00 patient records already in other studies. We plan to translate this model into the field of head injury and extract routine care data across a wide range of potential prognostic indicators (e.g. frailty, falls, co-morbidities) alongside automated imaging analysis. We have developed the CREAM Tea collaborative (Community Research Engagement in Ageing and the Mind) (14), which has run large events (60-100 attendees) and online focus groups. Feedback includes overwhelming support for this study and frustrations that health data isn t currently used to help predict risk in those most commonly affected. Comments included; Broad views on whether people would want neurosurgery depending on likely recovery, and therefore essential to estimate this on an individual basis and guide families with big decisions. Patients who engage with GP regularly feel this data could help doctors understand their baseline health No concerns about data being anonymously accessed to contribute to better understanding about recovery
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know