Subha Madhavan has worked 60-hour weeks regularly for the past 10 years, and she doesn’t show any sign of slowing down.

Madhavan splits her time between two main jobs — one as director of the Innovation Center for Biomedical Informatics at the Georgetown University Medical Center, and the other, more recently, as the chief data officer at Perthera, a McLean-based cancer therapy advisory service. She is also an associate professor of oncology at Georgetown, and leads a Georgetown-Howard University joint clinical research program for students.

How does she have time for it all? The Potomac resident said she gets in a couple of hours of work before her husband and 10-year-old son wake up in the morning, and another few after they go to sleep. She spends a few days a week at Georgetown and the remainder at Perthera. But when she’s on the move, she’s accessing medical databases on her iPhone, videoconferencing or advising her team over the phone.

Throughout her busy career, Madhavan has had one goal: to bring computer science into medical treatment.

Madhavan is a proponent of “precision therapy” — a medical approach that assumes patients are unique, and some are more likely to respond to treatment than others because of genetic or environmental factors. After analyzing large volumes of data about past patients, doctors can better advise current patients about which treatments are likely to work, she thinks.

At Perthera — a mashup of “personal” and “therapy” — Madhavan oversees the software-based processing of tissue samples. Founded in 2012, the approximately 10-person start-up advises patients with advanced cancer on treatments that might go beyond the standard options physicians might offer other patients. The company takes biopsies, analyzes the patient’s genes and proteins, and recommends to physicians a drug fit for the patient’s individual molecular make up.

The company, which processes biopsies over the course of a month, charges a fee for the service. (Eventually, the company hopes third-party payers and insurance companies will assume some of the cost.)

“I always wanted to translate my academic research into real patient care, and impact patient care, and saw this as a wonderful opportunity to make that transition,” Madhavan said.

She said she knew Perthera’s founders through research collaborations at Georgetown, and they asked her to join earlier this month.

When she’s not crunching data for Perthera, Madhavan is working on similar projects at Georgetown’s Lombardi Cancer Center. Madhavan oversees a 22-person team that is currently developing a mathematical model to predict which stage-two colorectal cancer patients are likely to benefit from chemotherapy, and which would likely relapse. Based on a 40-person sample size, the team’s model appears to predict relapse with about 95 percent accuracy, she said, though she noted the study needed to undergo several more trials to be considered scientifically valid.

“The case here is to find the right drug for the right patient,” Madhavan said. “It’s using information — it’s not the genes, it’s how the genes interact with the environment.”

Some medical systems use an oversimplified approach to determining which patients are eligible for which treatment, she said — if they have a certain condition, they are likely to need a certain treatment. “But it’s not a binary piece of information,” she said, noting other factors, such as if a patient is a smoker, could determine the treatment’s efficacy.

“It’s a huge challenge for oncologists to decide which ones to put on chemotherapy after surgery.”

In another project at the Lombardi Center, Madhavan’s team is predicting late effects — side effects occurring months or years after treatment — in pediatric cancer patients. Cancer patients as young as three- or four-years-old often experience seemingly unrelated symptoms later in life, such as memory disorders, reproductive problems, skin discoloration, among other symptoms.

Her team is using software to mine through the electronic health records of about 300 pediatric patients, searching clinicians’ notes about the child’s diagnosis, prescription dosage and any late effects. The software her team uses — much of which is open-source — has so far mined through about half a million patient notes in this study, she said, and uses a technique known as “natural language processing” to parse out clinician notes.

Madhavan, who earned her PhD in biological science at the Uniformed Services University of the Health Sciences in Bethesda, became interested in genes while studying hypoxia, a condition in which parts of the body are deprived of oxygen, especially among in gunshot victims.

“That was my initial foray into looking at the molecular biology of biomedical problems, and how understanding the genes and proteins could potentially solve clinical biomedical problems.”

Later, as an associate director for the National Cancer Institute’s Center for Biomedical Informatics, Madhavan led a team of scientists, physicians and software engineers to build a database of genetic data from brain tumor clinical trials, called “Rembrandt.”The project earned the team a “Service to America” medal in 2005.

It was as an undergraduate student in India during the 1980s that Madhavan first thought about combining computer science and biology, she said. Following in the footsteps of her father and grandfather — who had graduate degrees in chemistry and medicine, respectively — Madhavan earned bachelor degrees in chemistry and biology.

“We had this medical library, and when we had to access information, had to literally go in there and search the journal. I wondered if there was a better search technique that would get my information,” she said, or if computational software could be applied to biology.

“That really inspired me to connect the dots. Molecular biology and modern medicine do not exist without technology ... We need both to help reduce health-care costs, and it’s becoming a value-driven medicine.”

In the rare moments when she isn’t thinking about medicine and data science today, Madhavan attends Zumba classes and spends time with her son, an aspiring computer scientist.

And though it won’t be in the near future when she gets some free time, “I’ d love to learn the classical violin.”