Lui Sha is leading a team working on a best practice guidance system to help physicians eliminate preventable medical errors.
Lui Sha, the Donald B. Gillies Chair in Computer Science, understands the painful reality of preventable medical errors. To improve this all-too-common tragedy, Sha undertook a new challenge with the belief it will be computer science that provides a proper response.
Simply put, preventable medical error indicates that proper medical knowledge for treatment exists but was not followed. Preventable medical errors are a leading cause of deaths and a challenge to computer scientists.
The need to combat this problem grew as reports of a physician shortage have indicated that too few doctors will be available to address all of the needs of a growing, aging population. The problem has heightened as a deluge of medical information developed during the spread of COVID-19.
That’s why Sha now leads a group of researchers that includes fellow Illinois CS professor Grigore Rosu, Illinois CS postdoctoral researcher Maryam Rahmaniheris, Dr. Paul Jeziorczak of OSF Children’s Hospital of Illinois and Dr. Priti Jani of University of Chicago Medical School to develop a medical best practice guidance system for severely ill COVID-19 patients. This effort is funded by the C3.ai Digital Transformation Institute.
Their project, titled “COVID-19 Medical Best Practice Guidance System,” uses up-to-date medical guidelines expressed in precise software models. It will provide a guide for physicians to follow toward proper patient care for the disease.
“Preventable medical errors are painful for physicians, as it’s often caused from overwhelming workload and treatment of diseases outside of their expertise,” Sha said. “We want to help doctors care for their patients, which is especially difficult when a physician does not treat a complex disease in high frequency. This is like being a good driver, but then going to a new city only for someone to ask you to pick the best route."
“If we can say that GPS is important to driving, imagine how important a medical best practice guidance system like this can be to practicing physicians.”
By leaning on expertise of medical best practice guidance system projects with Carle Foundation Hospitals, OSF Children’s Hospital of Illinois, University of Chicago Medical School and, recently, University of California, San Diego medical school, Sha felt confident forming the work group and developing the COVID-19 guidance system.
The team then leveraged Rosu’s K Framework, which ensured a program written in mathematically sound methodology. Meanwhile, Rahmaniheris’ expertise in computational pathophysiology knocked down many barriers and provided strong connections with physicians.
“I can’t emphasize enough the capability of both Grigore and Maryam,” Sha said. “Maryam is exceptionally talented in the science of computational pathophysiology, which transforms complex medical knowledge in English to precise computing models. Also, Shuang Song is a talented PhD student who leads our software engineering efforts."
“Everyone involved helped form a great team, which made me confident that we could tackle any challenge ahead. I’m proud and fortunate to work with them all.”
Recently, Dr. Paul Jeziorczak of OSF Children’s Hospital of Illinois and Dr. Priti Jani of University of Chicago Medical School, signed on to provide versions of the model to help their physicians deliver the right care at the right time. Sha understands that COVID 19 has created unprecedented stress on medical staff. He believes this project cannot succeed without their medical expertise and devotion to patients’ well-being.
“Since 2000, when the Institute of Medicine wrote a paper called ‘To Err is Human,’ we’ve known that too much medical information exists for the medical community to manage,” Sha said. “When dealing with preventable medical errors, the first thing to do is to recognize this is not a medical knowledge problem. It’s about the misapplication of medical knowledge. That is why, computer science, can be a big part of the solution.”
Software quality is critical in the guidance system for acute care. Sha has asked Rosu to help with the formalization of guidance system software. Rosu’s K Framework could provide a rigorous, mathematically grounded system generating more confidence in the guidance system software.
“For the nurses and physicians utilizing the model, the K framework isn’t visible. To be honest, they probably don't care what programming language runs the tablet they hold in their hands,” Rosu said. “But what the K Framework does is eliminate their concern that the program might not do what the medical guidelines provide.”
With the technology built, the team needed one more significant piece to fall into place.
They needed to learn from expert physicians to optimize the treatment for a patient within the medical guidelines. This equates to the safe use of AI for acute care.
To account for this need, Thomas M. Siebel’s C3.ai DTI project came into play. Sha and his group needed medical connections and further funding to build staff in response to the challenge posed by COVID-19 beyond the seed funding for the proof of concept prototype. A common challenge faced in the use of AI for clinical research is how to share COVID-19 clinical data quickly. Looking ahead, a meeting will be scheduled between C3.ai, OSF Children’s Hospital of Illinois and University of Chicago Medical School.
“I can’t express enough that the dedication to this effort proves to me that C3.ai CEO Tom Siebel’s heart is in this effort,” Sha said. “He is a transformational leader with great compassion. There also is no better way for people to understand the impact of C3.ai than to produce a difference during this pandemic.”