Discover how medical professionals and MIT researchers are using data-enabled systems to help doctors around the world make the best healthcare decisions.
If you think it's hard to keep up with all the new software and hardware innovations, imagine doctors trying to stay abreast of medical advances.
"While wonderful new medical discoveries and innovations are in the news every day, doctors struggle with using information and techniques available right now," writes Leo Anthony Celi, assistant professor of medicine, Harvard Medical School, in the Conversation commentary Improving patient care by bridging the divide between doctors and data scientists. "As a practicing doctor, I deal with uncertainties and unanswered clinical questions all the time."
As to why there are uncertainties, Celi suggests the following reasons:
- Doctors are busy helping patients. So, there is precious little time to find and digest information in books and online.
- If there is time to wade through information, Celi feels more often than not the data is not specific enough to apply to a given patient or situation.
- A significant number of health records are still paper-based, making it difficult to disseminate vital diagnostic information.
Because of the above, Celi says, "We must work with what we carry in our heads, from personal experience and education."
Another constraint, perhaps even more important, is that the information available is usually not focused on the specific individual or situation at hand.
Enter big data
Celi feels there are opportunities for big data and information analytics in the healthcare field. "A digital system would collect and store as much clinical data as possible from as many patients as possible," writes Celi. "It could then use information from the past—such as blood pressure, blood sugar levels, heart rate, and other measurements of patients' body functions—to guide future doctors to the best diagnosis and treatment of similar patients."
To that end, Celi, who also serves as clinical research director of the Laboratory of Computational Physiology at Massachusetts Institute of Technology (MIT), along with fellow researchers, have been collecting patient data for the Medical Information Mart in Intensive Care (MIMIC). "It is a database containing information from 60,000 patient admissions to the intensive care units of the Beth Israel Deaconess Medical Center," writes Celi. To address concerns about security, Celi explains the data in MIMIC has been meticulously scoured so individual patients cannot be recognized.
Besides creating the database, Celi and the MIT group have taken it a step further. "We bring together front-line clinicians (such as nurses, pharmacists, and doctors) to identify questions they want to investigate, and data scientists to conduct the appropriate analyses of the MIMIC records," writes Celi. "This gives caregivers and patients the best individualized treatment options in the absence of a randomized control trial."
Not just for the United States
Celi is looking beyond the US, wanting to help doctors around the globe—in particular those in countries with limited healthcare resources—make better medical decisions through the use of data-enabled systems similar to MIMIC. "Often these countries have few or no medical records—even on paper—to analyze," explains Celi. "We can help them collect health data digitally, creating the potential to improve medical care for their populations."
Tap into local geniuses
As to how Celi intends to accomplish this, he and fellow researchers at MIT started Sana: an organization consisting of clinicians, engineers, policy, public health, and business experts, or what Celi calls local geniuses—professionals living in the countries with limited healthcare resources. Celi says it only makes sense to use local talent that is aware of the problems and how to resolve them.
Celi also emphasizes the need for openness. "Our approach is to democratize access to quality healthcare through open-source technologies, democratize knowledge through the exchange of learning across partners, and to democratize access to global networks of multidisciplinary experts," from the Sana website.
Rather than use computers, the people at Sana went straight to cellular technology, as cell phones are more prevalent in poor and rural areas.
As shown above, the cell phones collect, store, and transmit medical data to clinicians via the cellular network. Celi adds, "It can handle not only basic patient data such as height and weight, but also photos, X-rays, ultrasound videos, and electrical signals from a patient's brain (EEG) and heart (ECG)."
The next challenge was developing a training program. Keeping with the open-source theme, Celi and the people at Sana started training sessions (bootcamps) and collaborative workshops (hackathons). "In 2015, we held bootcamps and hackathons in Colombia, Uganda, Greece, and Mexico," writes Celi. "The bootcamps teach students in technical fields like computer science and engineering how to design and develop health apps that run on cell phones. Immediately following the bootcamp, the medical providers join the group and the hackathon begins."
The hackathon is where the local geniuses excel. "A hackathon brings people from different fields together over a short period of time to attack a specific problem or type of problem," explains Celi. "Attendees focus on a specific health problem, such as how to screen rural populations for heart disease or monitor children with epilepsy, using cell phones."
A final thought
From a doctor's perspective, the advantages of big data and its analysis are numerous and far reaching. Celi has made that abundantly clear.
- How big data can save lives by diagnosing healthcare failings (TechRepublic)
- The underexploited big data sweet spot for healthcare (TechRepublic)
- Big data will enhance healthcare, but to whose benefit? (TechRepublic)
- IBM Watson: The inside story of how the Jeopardy-winning supercomputer was born, and what it wants to do next (TechRepublic)
- Harvard medical professor: Big data and analytics help cure cancer (ZDNet)