Algorithm can tell if you have blood poisoning

Sepsis, or blood poisoning as it's commonly known, affects 40,000 people in Sweden every year, causing the death of approximately 8,000. This means that more people die in sepsis each year than in breast cancer, prostate cancer, and bowel cancer combined.

One reason why so many people die of sepsis is that the symptoms of sepsis are the same as for many mild conditions. In addition, it often has a very fast course. By the time the patient is correctly diagnosed, it is often too late to save his or her life. But soon an AI can help doctors make accurate diagnoses quicker.

Researchers at Uppsala University have used machine learning to develop an algorithm that can determine which patients are at risk for developing sepsis.

The algorithm, called Navoy Sepsis, takes into account about 20 parameters such as body temperature and results from blood samples from the patient. Navoy Sepsis then analyzes all parameters and calculates a figure that shows the likelihood that the patient develops sepsis.

The algorithm is in the final stages of clinical trials and will now be developed and marketed by the researchers' company Algodx. The plan is to commercialize Navoy Sepsis and make it available to hospitals next year.