Using AI to allocate patients to treatment can reduce hospitalisation

Using machine learning to allocate medical treatments during a pandemic can reduce the number of people to be hospitalised, a study has found.

The study, published in Jama Health Forum, looked at the use of machine learning to distribute medication, using the Covid-19 pandemic to test the model. The model showed that hospitalisations could be reduced by about 27% compared to actual and observed care. It worked by making sure that when medication was scarce, it was allocated to the patients most at risk.

Adit Ginde, MD, professor of emergency medicine at the University of Colorado and senior author on the study, said: “During the pandemic, the healthcare system was at a breaking point and many health care facilities relied on a first-come, first-serve or a patient’s health history to implement who received treatments.”

The researchers used machine learning to investigate how individual patients benefit differently from treatment. They found that the model they developed provided doctors, health systems and public health officials with more accurate information in real time than traditional allocation score models.

Mengli Xiao, an assistant professor in biostatistics and Informatics, who developed the AI-based mAb allocation system, said: “Existing allocation methods primarily target patients who have a high-risk profile for hospitalisations without treatments. They could overlook patients who benefit most from treatments. We developed a mAb allocation point system based on treatment effect heterogeneity estimates from machine learning. Our allocation prioritises patient characteristics associated with large causal treatment effects, seeking to optimise overall treatment benefits when resources are limited.”

The AI model used by the researchers incorporated a method based on Policy Learning Trees (PLTs). This method, designed to be used during periods of resource constraint, would optimise the allocation of monoclonal antibodies (mAbs) that neutralise the Covid-19 virus.

The PLT approach was able to help decide which treatments to assign to patients in a way that maximised the overall benefits for the population, thereby ensuring that those at the highest risk of hospitalisation are certain to receive treatments, especially when treatment is scarce. This is done by taking into account how different factors affect the effectiveness of the treatment.

The researchers hope their study will encourage public health bodies and policymakers to look into using machine-learning methods in future pandemics.

Doctors immerse themselves in the patient experience using VR

Another technology – this time virtual reality (VR) – has been helping doctors and other NHS health workers to immerse themselves in the patient experience.

The Insight Programme, which has been launched in Suffolk and Essex, uses virtual reality to enable staff to see the clinical experience from the perspective of patients. Wearing VR headsets, the workers are able to participate in interactive filmed scenarios in which they follow the patient’s journey from initial presentation and consultation to diagnosis and treatment.

The technology is being used by GPs, physiotherapists, paramedics and physician associates.

David Cargill, a GP at Stowhealth in Stowmarket, said he was “very excited” about the potential of the new training programme, which is being run at West Suffolk College.

He said: “We’ve had two focus groups evaluating the training modules for some time now and the results are very positive. It’s a totally immersive experience that is so much more effective than sitting in a lecture or a face-to-face tutorial. When you experience the training, you immediately feel empowered to put it into practice with a patient. It takes you through different scenarios and allows you to consult specialist advice on how to deal with the difficult questions patients often ask.”

The new training platform has been developed by the Suffolk and North East Essex Training Hub, the Suffolk-based Revolve Labs and Eastern Education Group.

The two-hour modules, delivered via special VR headsets or by using a laptop or mobile can be viewed in one go, or in bitesize 30-minute chapters.

The subject for the first module of the training programme is chronic pain. “It’s difficult having to explain to someone with fibromyalgia why they’re experiencing pain. Watching an expert do this is very empowering, particularly when combined with understanding from the patient’s point of view,” Cargill said.