Mobile Menu

10 MIT-IBM Watson AI Lab funded COVID-19 projects

Artificial intelligence, unsurprisingly, has been the leading technology pitched to solve most of our newfound pandemic problems, from finding a cure to rescuing the plummeting stock markets. Whilst AI could potentially have a transformative effect on many facets of COVID research, most of these systems lack sufficient investment and validation to produce results robust enough to be trusted.

To those ends, the MIT-IBM Watson AI Lab has chosen ten AI research projects to fund, which although broad in their focus, all aim to address the health and economic consequences of COVID-19 and shape our crisis response in future outbreaks.

The university-industry collaborative project, which was set-up in 2017 to develop AI hardware, software, and algorithms for real-world applications, funds about 50 projects per year. This recent announcement echoes the core ethos of the partnership; to advance AI’s potential for wider society.

Kim Martineau, writing for MIT News, outlines the ten research projects being funded:

1. Early detection of sepsis in COVID-19 patients

Clinical studies have shown that around 10% of COVID-19 patients develop sepsis and, for those that do, the mortality rate sits around 50%. Identification of at-risk patients can help hospitals prioritise care and resources for sicker patients. Daniela Rus, an MIT professor, is leading efforts to develop an ML system to detect signs of activated immune responses from patient blood samples.

2. Designing proteins to block SARS-COV-2

A project led by MIT professors Benedetto Marelli and Markus Buehler, is using a protein-folding method akin to one they previously developed to determine the structure of a honeybee silk protein that can be used as a natural food preservative. They are now focused on designing proteins that will block the novel virus from binding to host cell receptors

3. Saving lives whilst restarting the U.S. economy

MIT professors Daron Acemoglu, Simon Johnson and Asu Ozdaglar are using models to contrast variable lockdown strategies on their effects on the economy and public health.

4. Which materials work best for face masks?

Besides the much coveted, and seriously limited, N95 masks, the effectiveness of most other medical and DIY solutions remains unclear. Associate Professor at MIT, Lydia Bourouiba, and her team are developing methods and measurements to standardise mask evaluations across variable situations and weather conditions.

5. Treating COVID-19 with repurposed drugs

The Rafael Gomez-Bombarelli Lab are using 3D spatial data of the virus to determine if any already-approved drugs would be effective against the disease. They have employed the help of NASA’s and the U.S. Department of Energy’s supercomputers to accelerate screening.

6. A privacy-first approach to automated contact screening

MIT researchers Ronald Rivest and Daniel Weitzner, in collaboration with MIT Lincoln Laboratory, are using encrypted Bluetooth data from people’s smartphones to ensure that information remains anonymous and secure.

7. Overcoming manufacturing and supply hurdles to provide global access to a vaccine

The challenges of efficiently and equitably distributing billions of doses of a COVID-19 vaccine are entirely unprecedented. MIT professors, Anthony Sinskey and Stacy Springs are leading a team who are developing statistical models to evaluate manufacturing, supply considerations and trade-offs, in scaling up a vaccine candidate. They hope to give decision-makers evidence-backed guidelines on how to cost-effectively achieve global distribution.

8. Leveraging medical records to find a treatment

Using statistics, ML and simulated clinical trials, a project led by MIT professors Roy Welsch and Stan Finkelstein, will analyse millions of electronic health records, looking for signs that some already-approved therapeutics could be effective against COVID-19.

9. Finding better ways to treat COVID-19 patients on ventilators

IBM researchers Zach Shahn and Daby Sow, MIT researchers Li-Wei Lehman and Roger Mark, are developing an AI system to advise doctors of the optimal ventilator settings on which COVID-19 patients should be treated. Shorter durations on ventilation-assisted breathing can reduce the damage caused to the lungs while freeing up the limited number of machines. The data the team is using comes from patients in ICU with acute respiratory distress,

10. Returning to normal via targeted lockdowns, personalised treatments, and mass testing

MIT professor, Dimitris Bertsimas, is leading a team studying the effects of lockdowns on reducing infection rates and developing ML models to predict vulnerable patients and to recommend personalised treatments. Additionally, the group will develop an inexpensive, spectroscopy-based test for COVID-19 to deliver a mass testing method capable of producing readings within minutes.

More on these topics

AI / Covid-19 / ML

Share this article