Home / Blog / The computer says you’ll get well

The computer says you’ll get well

Neural networks analyse patient’s data as he is taken into ICU and calculate the death risk based on 20+ years worth of data. Credit: Annelaure B. Nielsen et al, University of Copenhagen

In the June issue of “The Lancet Digital Health” an article is reporting the study undertaken by several researchers and medical doctors  (mostly from the University of Copenhagen) on data from more than 230,000 Danish patients who were hospitalised over the last 20 years at ICU (Intensive Care Unit). Clearly patients ending up in ICU are in a critical condition and a few of them will not recover. The historical data, tracking patients outcome in 30 and 90 days, have been processed using artificial intelligence software, neural networks, to identify correlation and probability of recovery.

The software has been able to pick up what are the most relevant parameters, what seems to be the most promising approaches given a certain condition to increase the chances of recovery. Most interestingly, the software has been put to work on newly admitted patients. The software is fed with the patient data at admission and on those gathered in the first 24 hours. This provides enough information to make correlations with previous cases and to alert the doctors in charge on the risks and to suggest most -statistically- effective approaches.

The training of the neural network was done by feeding it with 85% of the available data to extract correlation and a model, then the model was tested on the remaining 15% of data. This is turns led to the refinement of the model, increasing the prediction accuracy of the system. Whenever available the system was also fed with historical medical data of the patient preceding his admission to the ICU, sometimes going back as much as 23 years. You can get all details from the paper, published under the Creative Commons license.

What I feel is important is that data are once more proving to be a game changer. In health care there are tons of data, most of them, unfortunately, are not accessible and I can understand the obvious privacy concerns, At the same time it is becoming more and more evident that if these data are made available we have the technology to leverage on them resulting both in an increased understanding of pathologies/effectiveness of cure and in the possibility to customise, finely tune, the cure to a specific person. I, for one, would be more than happy to share my medical data to the research community and to have my doctor use the results for increasing the effectiveness of cure.

I might be wrong, but I think that the majority of people share my same sentiment in this area. What we are missing is the general awareness that a digital transformation of healthcare would be extremely beneficial to the society as a whole (in economic sense) and to the single individual (better health care … and most importantly better health). In a way health care is the area where, given the technologies we have today, the digital transformation would be more effective and yet it is one of the areas that are lagging behind.

About Roberto Saracco

Roberto Saracco fell in love with technology and its implications long time ago. His background is in math and computer science. Until April 2017 he led the EIT Digital Italian Node and then was head of the Industrial Doctoral School of EIT Digital up to September 2018. Previously, up to December 2011 he was the Director of the Telecom Italia Future Centre in Venice, looking at the interplay of technology evolution, economics and society. At the turn of the century he led a World Bank-Infodev project to stimulate entrepreneurship in Latin America. He is a senior member of IEEE where he leads the Industry Advisory Board within the Future Directions Committee and co-chairs the Digital Reality Initiative. He teaches a Master course on Technology Forecasting and Market impact at the University of Trento. He has published over 100 papers in journals and magazines and 14 books.