Published by
Stanford Medicine


Big data

Big data, Clinical Trials, Events, Research

At TEDMED 2015: Benign drugs? Not under the lens of big data

At TEDMED 2015: Benign drugs? Not under the lens of big data

This year’s TEDMED was held Nov. 18-20 in Palm Springs, Calif. Stanford Medicine is a medical research institution partner of TEDMED, and a group of MD and PhD students who represented Stanford at the conference will be sharing their experiences here.

xCUEHR0MrJlqiC9phSMFFEjCxjrDDo54Bv0Hc18sYdkPicture this: you go to the doctor and find out that your cholesterol is high. Your doctor prescribes you a medication taken by millions of Americans for lowering cholesterol – Pravastatin. A few months later, you see your doctor again because of persistent depression, and again, you are given a commonly prescribed medication – Paxil.

Russ Altman, MD, PhD, opened his 2015 TEDMED talk with this seemingly innocuous scenario. But through the course of his talk, Altman demonstrated how his lab leveraged big data to reveal the adverse side effects of supposedly benign pharmacological interventions.

When choosing medications for my patients during my clinical rotations, I would often cite evidence from randomized controlled trials about the clinical benefits versus the risks of that particular drug. However, this evidence-based medicine has one major limitation: In clinical studies, patients are usually only on one drug.

My patients, on the other hand, would often come in with bags full of prescription bottles in order to show me which drugs they took, since there were too many medication names to memorize. Often, I found myself wondering quietly, “Is there any way to know if combining these drugs could lead to an adverse event?”

Continue Reading »

Big data, Events, Science

At TEDMED 2015: Using data to maximize human potential

This year’s TEDMED was held Nov. 18-20 in Palm Springs, Calif. Stanford Medicine is a medical research institution partner of TEDMED, and a group of MD and PhD students who represented Stanford at the conference will be sharing their experiences here. 

A tall, striking woman walked out onto the TEDMED stage. Energetically and confidently, Vivienne Ming, PhD, a theoretical neuroscientist and entrepreneur, went on to tell us about her passion for optimizing human potential utilizing big data.

There are many examples where Ming has used data for social good, such as matching refugees to their families with image-recognition software, but here at TEDMED she discussed personalized education. By looking for trends and patterns in classroom data, she has built an educational tool that predicts a child’s grade trajectory in the class and then chooses the highest impact personalized intervention for each individual student. Using data-science techniques to study society and education, Ming dazzles us with the possibilities of data in improving our capacity to reach our individual potentials.

Ming completely upended my view on data science in social issues

As a graduate student applying these same techniques to the field of genomics, I was intrigued. My understanding of data science applied to the social sciences, such as in the field of economics, was that the modeling remained fairly straightforward and simple: Understandable models were very important for the social sciences, and complex data-science models were not very interpretable. As a medical student, I found this frustrating – my interests lay both in cutting-edge computational advances as well as empathy for human suffering, and I couldn’t figure out how to apply my data-science skills beyond the world of science to answer questions of social and health inequalities.

Ming completely upended my view on data science in social issues. I was fortunate enough to have the chance to chat with her the next day, and I asked her about the computational work, as I was curious how complex her models really could be. She displayed a wonderful technical capacity and a deep understanding of how to choose the right algorithms for the social problems at hand. Granted, it’s still not common to find such imaginative interdisciplinary work combining cutting-edge computer science work and social science work. But Ming showed me a world of possibility around bringing data skills into improving everything from hiring to education to gender equality. I came away impressed, inspired, and excited about the possibilities of utilizing my own skills in the world beyond genomics.

Working at the intersection of two fields can be extremely challenging to impossible. And it’s particularly tricky to apply the latest data-science methods to societal questions. As such, to be able to intelligently and thoughtfully do the two together is an art – and Ming is a master of the important intersection where computation meets humanity.

Daniel Kim is a fifth-year MD/PhD student at Stanford. He studies biomedical informatics and genomics and is interested in all things data-related.

Big data, Cancer, Genetics, NIH, Precision health, Research, Stanford News

“Housekeeping” RNAs have important, and unsuspected, role in cancer prevention, study shows

"Housekeeping" RNAs have important, and unsuspected, role in cancer prevention, study shows

BroomsNot every character in a novel is a princess, a knight or a king. It’s the same for our cellular cast of characters. Most molecules spend their time completing the thousands of mundane tasks necessary to keep our cells humming smoothly. Many of these are referred to as “housekeeping” genes or proteins, and biologists tend to focus their attentions on other, more flashy players.

Now dermatologists Paul Khavari, MD, PhD, and Zurab Siprashvili, PhD, have found that a pair of housekeeping RNA molecules play an important role in cancer prevention. They published their findings yesterday in Nature Genetics.

As I explain in our release:

[The researchers] compared 5,473 tumor genomes with the genomes obtained from surrounding normal tissue in 21 different types of cancer. In many ways, cancer cells represent biology’s wild west. These cells divide rampantly in the absence of normal biological checkpoints, and, as a result, they mutate or even lose genes at much higher rate than normal. As errors accumulate in the genome, things go ever more haywire.

The researchers found that a pair of snoRNAs called SNORD50A/B had been deleted in 10 to 40 percent of tumors in 12 common human cancers, including skin, breast, ovarian, liver and lung. They also noted that breast cancer patients whose tumors had deleted SNORD50A/B, and skin cancer patients whose tumors made lower levels of the RNAs than normal tissue, were less likely than other similar patients to survive their disease.

The researchers used data from the National Institutes of Health’s The Cancer Genome Atlas to find that the RNAs are frequently deleted in tumor tissue. They further went on to show that the RNAs bind an important cancer-associated protein called KRAS and keep it from associating with an activating molecule.

“This is really last thing we would have expected,” said Khavari. “It was particularly surprising because my lab has been studying KRAS intensively for more than a decade, so it was quite a coincidence.”

The researchers believe that understanding more about how the RNAs inhibit KRAS activation could point to possible new therapies for many types of human cancers.

Previously: Listening in on the Ras pathway identifies new target for cancer therapySmoking gun or hit-and-run? How oncogenes make good cells go bad  and Linking cancer gene expression with survival rates, Stanford researchers bring “big data” into the clinic 
Photo by Rob Shenk

Big data, Biomed Bites, Research, Science

How one statistician is refining clinical trials

How one statistician is refining clinical trials

Biomed Bites is back. It’s a weekly feature that introduces readers to some of Stanford’s most innovative biomedical researchers. 

A love of mathematics propelled Ying Lu, PhD, to his current position as professor of health research and policy – with a specialty in biostatistics.

As the genomic revolution generates skyscrapers of data, biostatisticians like Lu are scrambling to develop more effective and efficient ways to design experiments and conduct clinical trials. Here’s Lu in the video above:

My research focuses on the development and application of novel, cutting-edge biostatistical methods in the evaluation, validation and comparison of new medical diagnoses as well as treatment interventions.

Lu also directs the nearby VA Cooperative Studies Program Coordinating Center, which conducts national clinical trials on topics such as stroke prevention, post-traumatic stress disorder and heart failure treatment.

Learn more about Stanford Medicine’s Biomedical Innovation Initiative and about other faculty leaders who are driving biomedical innovation here.

Previously: What I did this summer: Stanford medical student investigates health statistics and costs in Costa Rica, It all comes down to truth: Stanford med student digs in on public-health campaigns and How the FDA is promoting data sharing and transparency to support innovations in public health

Big data, Medical Apps, Patient Care, Precision health, Research, Stanford News, Technology

Precision health in practice: Using HealthKit to monitor patients’ blood-sugar levels

Precision health in practice: Using HealthKit to monitor patients' blood-sugar levels

Rajiv Kumar and patient - 560

Imagine having to keep track of your diabetic son’s constantly changing blood sugar levels by typing each individual reading into an email. Then, once in the doctor’s office, having to spend a chunk of your precious time with your clinician waiting for her to download that data.

That was the plight of Lori Atkins, whose son has Type 1 diabetes, until this March, when the Atkins joined a pilot project involving Apple’s HealthKit. Pediatric endocrinologist Rajiv Kumar, MD, is using HealthKit – a new technology that can securely share health data with third-party applications – to more easily monitor the blood-sugar levels of 10 patients.

A recent Inside Stanford Medicine article describes the project:

Patients like Blake wear a continuous glucose monitor that sends 288 blood-sugar readings a day to an Apple mobile device through Bluetooth. The data is securely transmitted via HealthKit into the patient’s electronic medical record at Stanford Children’s Health through the MyChart app.

The system also improves clinical outcomes, Kumar said: “Our endocrinologists are now able to easily assess large volumes of blood-sugar data between clinic visits — and quickly identify trends that could benefit from insulin dosing regimen changes.”

Kumar is planning to expand the use of the app to more of his patients.

Previously: A look at the MyHeart Counts app and the potential of mobile technologies to improve human health, Harnessing mobile health technologies to transform human health and A picture is worth a thousand words: Researchers use photos to see how Type 1 diabetes affects kids
Photo by Norbert von der Groeben

Big data, Patient Care, Stanford News, Technology

OrderRex taps decisions of thousands of “doctors like me”

OrderRex taps decisions of thousands of "doctors like me"

chen_vaAs a new clinician, Stanford’s Jonathan Chen, MD, PhD, struggled to treat patients with unfamiliar conditions. He yearned to ask one or, even better, dozens of more experienced physicians for advice.

For most people, that would be a passing wish. But not for Chen, who has a PhD in computer science and experience working as a software developer. (Oh yeah, he also started college when he was 13).

A recent article from the Center for Health Policy and Center for Primary Care and Outcomes Research (CHP/PCOR) describes Chen’s next steps:

“I thought about how the Amazon product-recommender algorithm works and thought, `Can we do this for medical decision-making?’” said the 34-year-old Chen, a VA Medical Informatics Fellow at Stanford Health Policy.

So instead of, other people who bought this book also liked this book, how about: Other doctors who ordered this CT scan also ordered this medication.

“What if there was that kind of algorithm available to me at the point of care?” he asked. “It doesn’t tell me the right or wrong answer, but I bet this would be really informative and help me make better decisions for my patients.”

Chen’s idea differs from the Green Button concept, which draws on thousands of medical records to search for patients with similar conditions. Instead, Chen is trying to capture doctor’s decision-making process by developing a digital platform to mine electronic medical records; he calls his project OrderRex.

It “looks for ‘doctors like me,’ and anticipates what the doctor wants before they ask for it,” Chen explains in the article.

Chen received a five-year National Institutes of Health grant and is working to develop OrderRex with the guidance of his mentor, bioinformatician Russ Altman, MD, PhD.

Previously: Push-button personalized treatment guidance for patients not covered by clinical-trial results, Big Data in Biomedicine panelists: Genomics’ future is bright, thanks to data-science tools and Euan Ashley discusses harnessing big data to drive innovation for a healthier world 
Photo by Joseph Matthews/VA Palo Alto

Big data, Genetics, In the News, Precision health, Research

Personal proteins: Assembling a “‘complete parts list’ of the human body”

3597686581_389d7b3df2_zGeneticist Michael Snyder, PhD, is on the forefront of a global effort to catalog — and investigate — the presence and activities of proteins in the human body. The worker bees of cells, proteins are responsible for the actions — such as germ fighting, digestion, reproduction and more — that keep us alive.

The task of tallying proteins is daunting, as a recent Nature article lays out:

Proteins… vary over time, changing during exercise, disease and menstrual cycles, for example. Another complication is that the most abundant protein can be about 10 billion times as common as the least.

Snyder started with himself and watched how his protein expression changed when he became ill with an infection. He also discovered his unexpected predisposition for diabetes. “I had no idea I’d turn out to be so interesting,” Snyder told Nature.

The piece outlines the multiple global efforts to “create a ‘complete parts list’ of the human body,'” as described by Gilbert Omenn, MD, PhD, head of the Human Proteome Project. Those endeavors, including the HPP, are using a variety of methods and tackling different tasks. For example, one is looking at proteins involved in disease, while another is systematically probing proteins produced by each chromosome.

Ultimately, Snyder said he hopes he and others can assemble protein inventories on as many as a million people. A key challenge of this work is what to do with, and how to analyze, the enormous amounts of data generated.

Previously: Gene regulation controls identity — and health, You say “protein interactions,” I say “mosh pit:” New insights on the dynamics of gene expression and ‘Omics’ profiling coming soon to a doctor’s office near you?
Image by Jer Thorp

Big data, Genetics, Research, Stanford News

Locking the door on big-data risks to privacy

Locking the door on big-data risks to privacy

Back doorUntil this week, you could have hacked into your rich Uncle Al’s account at a popular family tree website, downloaded his genome and then gotten your geneticist cousin, Todd, to help you find out if Al had a disease that could hopefully lead to an early and lucrative death. Thanks to a pair of researchers here, you won’t be able to do that.

Suyash Shringarpure, PhD, a postdoctoral scholar in genetics, and Carlos Bustamante, PhD, a professor of genetics, realized that an unnoticed back door to a network of genomic data sets was capable of revealing more about a person’s health than anyone would like. But thanks to the two men’s work, that back door will soon be locked tight.

In a new paper, published yesterday in The American Journal of Human Genetics, the researchers demonstrate both how someone might extract personal information from a major network of disease databases and how to prevent that from happening. As I explain in my story:

The Beacon Project has the potential to be enormously valuable to future genetic research… In their paper, the Stanford researchers suggest various approaches for making the information more secure, including banning anonymous researchers from querying the beacons; merging data sets to make it harder to identify the exact source of the data; requiring that users be approved; and limiting access in a beacon to a smaller region of the genome.

Their paper also bears importantly on the larger question of how to analyze mixtures of genomes, such as those from different people at a crime scene or the many different species of microbes in a person’s microbiome.

Previously: A conversation about the benefits and limitations of direct-to-consumer genetic tests
Image – of Back Door, Cliff Cottage watercolor painting – by Artistically

Big data, Clinical Trials, Health Policy, Precision health, Research

Push-button personalized treatment guidance for patients not covered by clinical-trial results

green buttonA pediatrician, a cardiologist and a biomedical informaticist walk into a pharmacy. They all look as if they could use some strong medicine. “We want a Green Button,” they tell the pharmacist in unison.

“Green Button? Hmmm. I can’t say I know how to compound that prescription,” the puzzled pharmacist replies. “But if all three of you are ordering it, maybe I should. Can you tell me what, specifically, goes into a Green Button?”

“A lot of patients,” reply the three thirsty health experts.

“OK, I’ll play along,” says the pharmacist, beginning to lose his patience. “What comes out?”

“If we knew the answer to that, we wouldn’t need a Green Button.”

Actually, that punch line is no joke. The “Green Button” signifies a profound, potentially pervasive approach that could revolutionize medical practice. In a just-published feature in Inside Stanford Medicine, I report on a futuristic (but not too futuristic) vision of a “learning health-care system” outlined in a 2014 Health Affairs paper by three Stanford experts: pediatric specialist Chris Longhurst, MD, cardiologist Bob Harrington, MD, and biomedical informaticist Nigam Shah, MBBS, PhD.

As I noted in that feature:

The randomized clinical trial is considered the gold standard of medical research. In a randomized clinical trial… participants are randomly assigned to one of two – or sometimes more – groups. One group gets the drug or the procedure being tested; the other is given a placebo or undergoes a sham procedure. … Once the trial’s active phase ends, rigorous statistical analysis determines whether the hypothesis, spelled out in advance of the trial, was fulfilled.

There’s one problem: Clinical trials select only a small, artificial subset of the real population. The rest of us are kind of out of luck.

“Clinical trials are designed to prove one thing,” Shah told me. “And you’re testing it on people with just one thing: type 2 diabetes, eczema, whatever. But most real-life people don’t have just one thing. They have three or four or five things.”

Enter the Green Button. Suppose you’re a clinician facing a patient for whom no clear clinical guidelines exist. Instead, according to the scheme depicted by Longhurst, Harrington and Shah, you press a virtual “green button” on a computer screen displaying your patient’s electronic medical record. This triggers a real-time search of millions, or tens or millions, of other electronic records. In a matter of minutes, up pops a succinct composite summary of the outcomes of 25 or 100 or perhaps 1,000 patients very similar to the one in front of you – same race, same height, same age, same symptoms, similar medical histories, lookalike lab-test results – who were given various medications or procedures for the condition you’re hoping to treat. Those “lookalikes,” it turns out, respond much better to one treatment than to the others – something you’d have been hard put to guess on your own.

That’s all very nice, you say. Now I get your “artisanal faux-joke” lead. But, you ask, why does the button have to be green? And I answer: It doesn’t. But the other good colors were already taken.

Previously: Widely prescribed heartburn drugs may heighten heart-attack risk, New research scrutinizes off-label drug use and A new view of patient data: Using electronic medical records to guide treatment
Photo by Green Mamba :)–<

Big data, Events, Precision health, Stanford News

Sino-U.S. Symposium brings researchers to Stanford to discuss precision health, big data

Sino-U.S. Symposium brings researchers to Stanford to discuss precision health, big data

LSINO-US panelistsast week, more than 300 health researchers from China and the United States converged at Stanford for the ninth Sino-U.S. Symposium on Medicine in the 21st Century. At this two-day event, health experts, thought leaders and entrepreneurs, including Lloyd Minor, MD, dean of the School of Medicine, and Jerry Yang, the Taiwanese-born co-founder of Yahoo, shared their knowledge of genomics, medical apps, and other topics related to this year’s theme: Big data in health care.

Minor kicked the symposium off saying, “We have the opportunity to harness the power of genomic data and electronic medical records, and to deliver better care, more personalized care for acute illness and, perhaps even more importantly, to predict and prevent disease before it even occurs — thereby moving the focus of medicine from sick care firmly toward health care.”

My colleague describes highlights from the event, including a discussion of how mobile devices can play a larger role in health care, in an online news story:

In China, clinics are so crowded that people line up in the morning to get a lottery number to be seen, [Alan Yeung, MD, professor of cardiovascular medicine] said. Yet, 1.3 billion people there own a smartphone that can potentially help monitor health. Globally, he said, 4.8 billion people own a cellphone.

“We could score someone’s risk of a heart attack and, depending on their risk factors, give them medications that would lower their risk,” said Yeung. “The idea at the end of the day is instead of one patient coming to a clinic, health-care providers come to a small clean room to monitor tens of thousands of patients and see who is in trouble.”

Cloud computing that was monitoring people’s heart rate, heart rhythm, blood pressure and glucose levels, for example, could light up when heart attack risk factors started to shoot up for a particular person. “We could schedule a quick call and find out what’s up,” said Yeung, “and then change whatever the problem is before they become entrenched in their habits.”

Previously: A conversation on the promises and challenges of precision healthHow Stanford Medicine will “develop, define and lead the field of precision health”At Big Data in Biomedicine, Stanford’s Lloyd Minor focuses on precision health and A look at the MyHeart Counts app and the potential of mobile technologies to improve human health
Photo of event panelists by Norbert von der Groeben

Stanford Medicine Resources: