The concept of “personalized medicine” gets thrown around a lot these days. And as an oncologist and a palliative care doctor, who advocates for using data to get information that can provide more personalized cancer care, I believe strongly in it. But what would truly personalized oncology look like?

More than 14 million people are living with cancer in the United States, according to the National Cancer Institute — and some 1.7 million people are expected to be diagnosed this year — yet researchers and doctors learn from only a tiny percentage of them. Clinical trials represent just three percent of cancer patients. Meanwhile, vast amounts of data on the millions of other people being treated for cancer — what drugs they took, which therapies worked, how long they lived, what other diseases they suffered from, etc. — is locked away in electronic health records, disorganized databases, doctors’ notes and the minds of patients and doctors.

Janet is a woman with late-stage melanoma who was in my care. She had hoped to use the data about all the tests she’d taken and treatments she’d undergone to learn more about the deadly disease she faced and how to treat it. But the hurdles to the sharing and analysis of data were too great, then, and she died before personalized medicine could take shape.

While roadblocks remain with how to share data between hospitals and health centers, as do other thorny issues related to privacy and regulation, which I’ll look at in the next post, it’s now possible to see where personalized oncology is headed. And oncology, because of cancer’s complexity and the sheer number of people affected by it, is something of a test case for personalized medicine more broadly.

At its core, personalized oncology would systematically organize and analyze all of that data about patients, and aggregate it with information about oncology research, to provide evidence-based medicine that is more accurate than currently possible. The database would include information from the patients themselves, clinical and administrative data, molecular and biological information, and research data, including from clinical trials. It would ask doctors clarifying questions when needed, to fill in an information gap or refine understanding. With each piece of data entered, the database would become smarter, constantly evolving as a real-time resource that doctors could tap into.

While people have hoped that computers would transform medicine since the 1960s, a number of pieces — from the requirements for electronic medical records and the spread of handheld devices to the proliferation of precise laboratory tests foretelling a tumor’s biology — are coming together now to create the conditions for personalized oncology. Both private-sector and public-sector efforts are currently being built.

The good news in oncology is that the number of cancer treatments has exploded over the past decade, with new innovations like immuno-oncology drugs and lock-and-key matched designer cancer treatments. And with biotech’s expansion, many of these newer therapies include biomarkers. Yet our ability as oncologists to sort through them and to identify which ones will work for which person and which tumor, remains rudimentary at best. At the same time, patients are more than just a compilation of biomarkers: They have their own stresses and concerns, their own values and desires, which make the best treatment for one not quite so good a fit for another.

Put this all together and where we are now is that when you’ve got a serious health problem, your doctor needs to amalgamate a lot of disparate pieces of information on the fly to take care of you.

The flood of information spans personal details (How well do your kidneys work? What’s your family history?) and research information (What treatments worked in similar situations in clinical trials? Are there new drugs that look promising for situations such as yours?), among them. In addition, hopefully, some of that data is also about your personal concerns and values (Will you accept blood transfusions? What’s your goal for treatment?).

All of these pieces of information, published and unpublished, need to be pulled together to come up with a treatment plan. Even for the best oncologists, the complexity has become overwhelming, something that technology can help alleviate.

But there’s another aspect to personalized medicine that often gets overlooked. When others talk about personalized medicine, they often focus on the variety of tests that can show the biologic basis of the disease in question. Is a breast cancer positive for the HER2 gene, or negative for it? If the former, the patient can be treated with Herceptin; if negative, the drug will be worthless. These lab tests are a crucial piece of information, but they are only one piece in a sea of information that needs to be understood.

Personalized oncology goes far beyond the question of one test,  or one lab result, and choosing one drug. In a field in which no two cases, and no two patients, are exactly the same, personalized oncology means understanding the differences between treatment regimens for men and women, and for patients in their 80s versus those in their 40s, as well as making decisions based on the genomic information contained in someone’s specific tumor.

Consider the case of Joe, an older man with lung cancer, who came to my clinic. He was so short of breath that he couldn’t even raise his arms to hold up a newspaper, and everyone in the hospital was trying to figure out what to do to help him.

He was getting chemotherapy, and a whole host of medications, and he’d already had his lungs drained. Yet every day, about 3 p.m., he would get worse. And every day, his team of doctors would huddle around, poring over his chart, and sometimes changing his steroid dosage.

After one of the doctors questioned whether the problem was related to his physical therapy appointment half-an-hour earlier, someone went down the hall to talk to his physical therapist. “Don’t you know that his son comes in every day at 3:30,” she said, “and his wife and son hate each other, and he doesn’t know what’s going to happen to the son when he dies.” And just like that the assumptions about what was going (and how wonderful it was that his son came to visit) were up-ended: No one had considered the dynamic between Joe’s wife — who was his second wife, and younger — and his son.

We as oncologists are always searching for new treatments and new technologies that will stave off a deadly disease. But sometimes in this frenetic search for an answer, we miss the most obvious issues. For Joe, no tube or new drug dosage was going to address the problem. And so he spent nine days in the hospital, and no one asked him what was going on. Personalized medicine means using all of the data, and that includes patients’ personal stories.

The technology itself is a tool, and we’re on the cusp of creating these massive databases that have the potential to transform medicine. But it’s important to remember that the goal is to leverage all of that data to empower patients and provide better care. Personalized healthcare means putting people in the center — their stories, their data, their needs — informed by science and history.

Amy Abernethy is a hematologist-oncologist. This article originally appeared in athenahealth’s Health Care Leadership Forum.