Homegrown TMU expert gives big data a human face:
Alumnus joins faculty, leads team to “top geeks” award
Usman Iqbal arrived at TMU from Pakistan in 2010 expecting to get a master’s degree. Instead he got two degrees, important roles editing three scholarly journals, more than fifty scholarly publications in six years – and now a tenure-track position with the Global Health and Development program.
But this story isn’t about what TMU has done for Dr. Iqbal; it’s about what he’s doing for TMU’s global reputation with his research.
On 4 and 5 March 2017, his team took top honors at the University of Melbourne’s Datathon, a short-term but high-pressure competition that started at the Massachusetts Institute of Technology (MIT). This Australian award was based on finding promising patterns in a given data set, and TMU’s six were competing in a field of 120 participants.
But what makes such studies qualify as big data? “Velocity” was the word he chose, using the examples of Twitter (daily volume: 90 million tweets) and Facebook (30 billion posts per day). Dr. Iqbal then offered some analogies:
“How would you go about building a fire engine out of Lego toy blocks?” His answer is that you would study a variety of different information sources to try to approximate the truck’s various functions.
“How would you extrapolate a coloring book’s content if you had only one page, or design a puzzle if you had only one piece to work from?” You would look for patterns – and that’s exactly what big-data studies focus on. Not causes or theories, just promising correlations.
Cheaper, faster, better results
If this sounds superficial, consider his sobering prediction that managers will soon be replaced by such reports noting metrics and inefficiencies. Computers will replace many other job categories as human expertise becomes too expensive, too slow and less reliable – witness the rise of computerized accounting and legal services.
But TMU’s business is health care. Here too, apps are proving irresistible because they cut costs while ensuring consistent quality.
Dr. Iqbal noted that Taiwan’s single-payer system is second only to Denmark’s in providing efficient, high-quality universal health services. But another of its virtues is the information that it provides through various databases.
In particular, each reimbursement claim lists the recipient’s address, diagnosis, medications, age, coexisting conditions, length of hospital stay, mortality, even income level. Other databases can interlink to compare various categories, for example relating these doctor’s office visits to information on air quality on various days, or noting spending patterns in the casualties and services after catastrophic events such as earthquakes.
That is why he said datathons are as demanding and exciting as hackathons (which Dr. Iqbal also excels in). His team of two staff and four students had two days to compete with other groups to find the most interesting and useful patterns in the data presented to them by contest organizers.
“They give the data, we find the questions: ‘Is it important? Is it doable?’” Dr. Iqbal explained. In Melbourne, they took electrocardiogram data and looked for heart rate variations that could predict cardiac arrest. Such information allows better preventive care by identifying higher-risk patients for individualized treatments.
“We were going to try to predict intensive care unit infections” based on the same data, he said. Many of the 120 participants will meet again at upcoming datathons planned in Malaysia and, later this year, at TMU.
Big data shakes up the drug world
Taipei Medical University has years of collaboration with MIT on M-health – health applications involving mobile phones. Dr. Iqbal clarified that these tend to be service- and practice-oriented, rather than research-oriented as datathons are.
One productive research area is long-term risks of various medications. The entire medical world has heard echoes of TMU’s findings regarding benzodiazapines, a common family of medications used for sleep. TMU found that those who took these drugs long-term had higher cancer rates.
Finding reasons is not as easy as finding correlations, Dr. Iqbal cautioned. He noted that the news about some drugs is good: aspirin use is associated with lowered rates of pancreatic cancer, for example.
“Drug research generally stops when medications hit the market,” he said. With this termination of evaluation, it is difficult to track long-term effects of new drugs, especially regarding cancers and other risks. Big data makes it easy and cheap to find this information — without worrying about human subjects and ethics board clearance.
“You have to design the methodology very carefully,” Dr. Iqbal said. But in translational medicine, it’s helpful in finding useful combinations of factors or drugs. And it’s very efficient, he said: “Big data reduces the time and money needed for the same research results.”
Up through the ranks
Dr. Iqbal arrived at TMU in 2010 on a scholarship for master’s degree studies in Health Care Administration. He hadn’t planned to stay for his Ph.D. studies, and quite nearly went to Sydney instead.
“I liked the environment here,” Dr. Iqbal recalled. In his final semester he had gotten involved with Dean (of College of Medical Science and Technology) Jack Li’s Biomedical Informatics lab and played a role in the SANA project – a global health informatics initiative – so he decided to stay.
“There’s more variety and range of things to do here,” he said. Studies elsewhere seemed narrow by contrast, as students arrive with a thesis proposal and are told to “stay within that fence.” Here he became involved with a variety of research leading to dozens of publications, as well as with scholarly journal publishing.
“Here I can taste all the flavors,” he said with a smile. So despite offers with better pay (one was from Spain), Dr. Iqbal stayed after his Ph.D. to take a faculty position last August. He said part of the reason was gratitude.
“Taiwan is open to international students; it’s friendly and safe. The world knows Taiwan is good at research, so graduates are highly employable around the globe. There’s good accommodation for foreigners at TMU too.”
This opportunity comes at a price – the infamous Taiwan work ethic. “In other countries you aren’t expected to work after business hours, but in Taiwan the popular social media apps mean that business communication never stops. It helps that I don’t mind long hours and am pretty focused on my work.”
As a first-year doctoral student, his paper on benzodiazapines and pneumonia was published in BMJ Thorax, a leading journal with an 8+ impact factor. This is the kind of opportunity that made a TMU education into a life at TMU – and both sides are grateful that Dr. Iqbal chose to continue his work here.
For interviews or a copy of the paper, contact Office of Global Engagement via global.initiatives@tmu.edu.tw.