Sepsis is one of those ugly conditions that probably doesn’t get the attention it deserves, but that isn’t to say that there haven’t been innovations emerging to tackle it (many of which I’ve covered here).  The latest of these comes from a team at Imperial College London, who have developed an AI-based tool to predict the best treatment plan for patients.

The work, which was documented in a recently published paper, analyzed patient records from around 100,000 individuals from 130 intensive care units over a 15 year period, including the decisions made by their doctors in each case.

The system, which is known as AI Clinician, was able to accurately predict the best treatment plan for patients, with the technology often able to do so more effectively than human doctors could.  It’s an outcome that the team hope will result in AI Clinician being deployed in intensive care units across the UK.

“Sepsis is one of the biggest killers in the UK – and claims six million lives worldwide – so we desperately need new tools at our disposal to help patients. At Imperial, we believe that AI for Healthcare is the solution. Our new AI system was able to analyse a patient’s data – such as blood pressure and heart rate – and decide the best treatment strategy. We found that when the doctor’s treatment decision matched what the AI system recommended, they had a better chance of survival,” the team say.

Treating sepsis

Sepsis often results in a significant fall in blood pressure, which can have a tremendous impact on the organs of the patient, which can in turn have tragic consequences.  To mitigate this risk, doctors often give extra fluids alongside medication designed to tighten the blood vessels and therefore raise blood pressure. Whilst this response is fairly standard, there remains considerable debate around how much fluid to give, and when to start the medication.  Clinical guidelines exist for both, but they only really provide general advice.

The AI system was able to go beyond this general advice and provide specific recommendations for each patient at their unique moment in time.  The system saw records from nearly 100,000 patients, which is over 6 times more than the average intensive care doctor will see in their lifetime.

“An intensive care doctor will see roughly 15,000 patients by the time they retire. Yet this system has seen nearly 100,000 patients, it has the life time experience of 8 doctors, and has learned from each of those cases what the best decisions were for each situation,” the researchers say.

This not only enabled the system to be effective in common cases, but also in rarer cases where human doctors may lack experience and exposure. The results suggest that when the recommendation of the doctor and AI varied, it usually did so in the sense that the doctor would prescribe too much fluid and too little medication, although this was a general trend that did vary for each patient.  The team believe that AI Clinician could be an invaluable decision support tool for clinicians however, and they plan to test the system out in the UK health system. The work is another example of what can emerge when AI technologies have access to large volumes of high quality healthcare data, and it will be fascinating to see how the project evolves once it’s unleashed on the notoriously impregnable National Health Service.