Technical problems with large medical data: gathering information

According to the June 8 news, Science and Technology blog website Recode.net published today that researchers in many institutions are exploring ways to use artificial intelligence and large data to find better treatments for difficult illnesses such as cancer, but they are having trouble collecting patient information.

The main contents of the article are as follows:

Researchers at IBM, Boston Pharmaceutical company Berg Pharma, Memorial Sloan-Caitlin Cancer Center (Memorial Sloan Kettering), UC Berkeley (ca Berkeley) and other institutions are exploring how to use artificial intelligence and large data Find better ways to treat disease.

But the biggest challenge in making the most of the computing tools in the medical world is that the vast amount of data is shelved-or not digitized from the start.

Early medical research or patient records are often locked in a medical company's file or in a cabinet in a doctor's office.

Patient privacy issues, conflict of interest between companies, and the sheer lack of electronic medical records impede the sharing of information in the medical field, making each treatment an isolated event. If information sharing in the medical field can make progress, there is a good chance of finding more general treatment options.

Michael Keze Michael Keiser, a lecturer at the University of California, San Francisco, points out that when you can analyze clinical trial data, genomic data, and electronic medical records for 100,000 patients, you will be able to discover treatments that you could not find in the past, compared to the information that you had previously been exposed to a few patients.

Given this prospect, some organizations are starting to integrate medical data.

At the end of last year, the American Society of Clinical Oncology (Anglo powering of clinical Oncology,asco) announced the initial progress of its "CANCERLINQ" project. "CancerLinQ" is a "fast learning system" that allows researchers to access, access, and analyze the medical records of anonymous cancer patients.

In April this year, a nonprofit group with a large number of major pharmaceutical companies, the CEO of the Cancer Life Sciences Association (Roundtable on Cancer), announced the launch of the PDS Program (Project Data Sphere). The plan will build a platform for data sharing and analysis of the third phase of cancer clinical trials, with the initial dataset being provided by AstraZeneca, Bayer, Celgene, Memorial Sloan-Caitlin Cancer Center, Pfizer, Sanofi, etc.

These data have been removed from patients ' personal information and are uniformly numbered for free use by life sciences companies, hospitals, medical institutions and independent researchers. They can access the analysis tools built into the platform or insert data into their software.

Martin Murphy, chief executive of Cancer CEO Roundtable, said the PDS program could help Martin Melfi the discovery of little-known cancer-candidate drugs that might have some effect on some cancers. In a particular study, these drugs may be discarded because they do not meet the main goals of the study.

Other efforts to advance information-sharing in the medical field include the Global Genomics and Health Alliance (Alliance for Genomics and Tiyatien) and the European Institute for Biological Information (EMBL-EBI), which is organized from multiple medical institutions, research universities, life sciences companies, etc. Maintenance of the molecular biological database, and the United States National Institutes of Health (Cato of Tiyatien, NIH), the establishment of the "Biological labeling community (biomarker Consortium)" and so on.

At the same time, Flatiron Tiyatien, a start-up company with large data services for oncology, completed a 130 million dollar B-round financing last month, which was led by Google's VC venture, Google Ventures. Flatiron Tiyatien has created an "oncology Yunping (Oncologycloud)" To extract and integrate clinical data and medical cost data from patient electronic medical records (EMR).

The system makes it meaningful to retain data in both the doctor's office and the hospital in an unsustainable and unstructured format, thus enabling the analysis of treatment for large groups of cancer patients. Ideally, it can find out which treatment is effective for which types of cancer patients.

"Flatiron Tiyatien focus on the clinical data of so-called ' real world ' patients," said Nat Turner, Flatiron Tiyatien co-founder Nait Tena. In the United States, only 4% of cancer patients are involved in prospective clinical trials, so we are trying to extract and integrate data from the remaining 96% patients. ”

He said: "To really understand what is effective for cancer, what other patients are receiving treatment, and what the results of research in the field of cancer, the relevant agencies should open" to identify (de-identified) medical data and anonymous cases, this is Flatiron Part of the Tiyatien vision. ”

Privacy risk

To be sure, the openness of medical information should be very cautious. Medical information is highly sensitive, so any privacy risk requirement should be considered carefully.

The extent to which medical information can be opened depends on concessions made by the whole society. Many people firmly hold the view that saving lives is the most important. But social change is not fast enough, influenced by old habits and outdated rules, David David Patterson, a computer science professor at the University of California, Berkeley. Patterson is dedicated to machine learning tools for cancer research.

"For researchers in the computer field, we are accustomed to Internet time and Moore's law," he said. But now we can't get the authorities to agree that it's very frustrating for us to have a lot of fast food to collect data and to integrate. ”

"Patient privacy is important, but making progress in the field of cancer treatment is equally important," he said. The advantage of bringing a lot of treatment information together is that we can make progress in overcoming this terrible disease. ”

None of the experts interviewed has been able to cite any breakthroughs in cancer treatment such as big data and so far. After all, these technologies are new, and medical data sets are just being integrated, and clinical trials can take years.

But almost everyone agrees that researchers are on the verge of a major breakthrough in cancer treatment.

If the cancer is likened to a mountain, Murphy said, it is now nearing the edge of the summit, an unprecedented height.