3 New Machine Learning Developments in Healthcare

Machine Learning accelerates the healthcare sector in a cost-effective manner

Machine learning

Machine learning

With advancing technology, machine learning is witnessing rapid development in all sectors. While coming to healthcare it has majorly impacted the system to a great extent. ML in healthcare can do wonders by processing huge datasets beyond human capacity, it can convert analysis of the data into clinical insights that can aid physicians providing care leading to better outcomes. Let’s look at the 3 new machine learning developments in the healthcare sector.


Predicting COVID-19 severity 

The National Covid-19 Cohort Collaborative (N3C) is a centralized electronic health record that has a Covid-19 cohort to date. With this data, researchers found that Treatment with blood thinners may reduce death in Covid-19 patients. 

This data can support rigorous proof-based development of predictive and diagnostic tools that can inform clinical care and policies told the team of researchers from the Universities of Colorado, Rochester Medical Center, Michigan, and John Hopkins. The research has used data from 34 medical centers that include 174,568 adults who tested positive for Covid-19 and 1,133,848 who tested negative between January 2020 and December 2020. 

“The machine learning models accurately predicted ultimate clinical severity using commonly collected clinical data from the first 24 hours of a hospital admission”, says tellen D. Bennett from the Department of Pediatrics at Colorado’s School of Medicine. The most prominent predictors in machine learning are the age of the patient, signs, and laboratory values.


Stroke recovery prediction 

A team of international scientists led by EPFL has developed a system that can combine information from the brain’s connectome and machine learning for assessing and predicting the outcome of stroke victims. For this particular study, scientists have analyzed connectomes of 92 patients 14days post-stroke. They tracked changes up to three months later accessing the motor impairments with a standard scale. This helped them to monitor the changes in the patients while they were recovering. 

The collected information was put into a ‘support-vector machine’ or SVM that was a type of machine learning model which maps an input onto an output. The SVM was beforehand trained between patients with natural recovery from those without whole-brain structural connection. This machine learning system was able to identify neuronal network patterns to make accurate predictions on outcomes of stroke patients. 

A neuroscientist and director of the Defitech Chair for Clinical Neuroengineering at EPFL’s School of Life Sciences, Professor Friedhelm Hummel said, “This tool can support the prediction of individual courses of recovery early on and will have an impact on clinical management, translational research, and treatment choice”.


Detecting Alzheimer’s during phone conversations

Alzheimer’s is a disease with no cure until now. Millions of people suffer from this disease worldwide. And so scientists have started to research more on new ways to predict who is likely to be affected with the disease. Earlier researchers show that some of the early signs of Alzheimer’s include speaking more slowly than normal. While other research is still in progress to recognize such speech difficulty. The team in Japan has been using the Telephone Interview for Cognitive Status (TICS-J) test where the conversations that take place on phone are recorded and studied. Since it is a little hectic to listen to conversations manually, researchers have designed a machine-learning algorithm to listen and analyze phone conversations. They suggest that these algorithms could be used to provide a cost-effective and accessible form of early Alzheimer’s testing. 

This is how machine learning is accelerating new developments in the healthcare sector.