A recent study led by researchers at the Perelman School of Medicine at the University of Pennsylvania has suggested that artificial intelligence (AI) can significantly enhance patient care by analysing data from multiple hospitals to create more tailored patient profiles. This research, published in the journal Cell Patterns, highlights the importance of understanding and addressing the specific needs of varied patient populations across different healthcare facilities in the United States.
The study focused on long-COVID patients, drawing on electronic health records from eight paediatric hospitals. By employing a machine learning technique known as "latent transfer learning," the research team managed to identify four distinct sub-populations among long-COVID patients. These groups included individuals with:
- Mental health conditions, such as anxiety, depression, neurodevelopmental disorders, and attention deficit hyperactivity disorder.
- Atopic and allergic chronic conditions, including asthma or allergies.
- Non-complex chronic conditions, like vision problems or insomnia.
- Complex chronic conditions, encompassing heart or neuromuscular disorders.
Yong Chen, PhD, who is a professor of Biostatistics and the senior author of the study, indicated the limitations of previous studies. He remarked, "Existing studies pool data from multiple hospitals but fail to consider differences in patient populations, and that limits the ability to apply findings to local decision-making." The research thus aims to provide a model that offers broad insights while allowing for precise applications within individual hospitals.
By pinpointing the distinct needs of these sub-populations, the study’s authors posited that healthcare providers could avoid a blanket treatment approach that may overlook the complexities of patients with higher risks. Qiong Wu, PhD, the study’s lead author, explained, "Without identifying these distinct subpopulations, clinicians and hospitals would likely provide a one-size-fits-all approach to follow-up care and treatment." The study highlights that patients with complex chronic conditions were found to have the most pronounced increase in hospital admissions and emergency visits.
The researchers assert that had the latent transfer learning system been deployed at the onset of the COVID-19 pandemic, it might have facilitated better resource allocation, potentially aiding hospitals in predicting the need for ICU beds, ventilators, or specialised staff. Wu noted, "This would have allowed each hospital to better anticipate needs for ICU beds, ventilators, or specialized staff-helping to balance resources between COVID-19 care and other essential services."
Looking beyond the acute crises of the pandemic, Wu also pointed out that the AI system could be beneficial in managing more common chronic conditions, such as diabetes, heart disease, and asthma, which often have varying prevalence and treatment requirements depending on regional health factors and the resources of the hospitals involved.
The team is optimistic about the practical application of their findings, suggesting that this system could be introduced at numerous hospitals with a reasonably straightforward data-sharing infrastructure. Wu stated, "By utilizing the shared findings from networked hospitals, it would allow them to gain valuable insights."
The research was supported by numerous grants from the National Institutes of Health and the Patient-Centered Outcomes Research Institute.
Source: Noah Wire Services