Diseases are treated based on the knowledge of statistics from large populations. For example, hospitals treat patients in different wards or buildings based on their disease. Although patients are treated as a group, they respond – or don't respond – as individuals. In fact, many patients go through trial-and-error until the right medication or dosage is found.
Businessweek noted that of the hundreds of billions of dollars spent on prescription drugs, over 40 percent went to medications that didn't help the patient. And billions more are being spent to treat adverse drug reactions and complications resulting from the medications themselves.
Healthcare professionals are responding to these error-prone and wasteful treatments by exploring "Real World Evidence." It's a term widely used within the medical field for collecting and reviewing the impact of chronic disease treatments to find the most-effective options for personalized care. At last week's Clinical Genomic Analysis workshop in Haifa, Israel, IBM Research – Haifa and the Edmond J. Safra Institute at Tel Aviv University hosted scientists and physicians working on bioinformatics research being conducted to do exactly this: offer "Real World Evidence."
|LTR: Rick Kaplan, Oded Cohn,|
"The gold mine of information available to the various stakeholders in the pharmaceutical and medical industry is just waiting to be tapped," said Michal Rosen-Zvi, manager of clinical genomic analytics at IBM Research – Haifa and organizer of the workshop.
"This new trend is very much in line with the rapid development of machine learning analytics and data-mining technologies to extract insight from masses of data," Rosen-Zvi said.
"Companies can now collect information that ensures their responsibilities go beyond pre-clinical and clinical trials, and use the data collected afterwards to optimize drug usage, efficacy, pricing, and security." This provides a situation in which patients can avoid complications and pharmaceutical companies can improve business by better targeting the drugs.
Connecting genes to treatment
Gabi Barbash, director general of the Tel Aviv Sourasky Medical Center (Ichilov) spoke about the inefficient drug therapy for cancer as a motivator for personalized medicine, and the new directions in genomic-based cancer therapy. He dispelled the myth that all mistakes in gene coding cause disease. In reality, only five percent of the gene coding contains abnormalities that cause disease. These abnormalities are found in some of the gene's single nucleotide polymorphisms – or SNP (pronounced "snip").
|Prof. Gabi Barabash|
"Not all SNPs have diagnostic implications, but by correlating the SNPs and the disease, we can find out which genes are linked to which diseases," Barbash said.
"By comparing the genome of people suffering from a certain disease with the genome of those who don't, we can identify the SNP involved – and then try to find out whether the SNP is the cause of the disease, or whether the disease itself has changed the SNP." He sees these connections as the basics of the genome-wide studies that can help improve treatments.
Watson, IBM's question answering machine that understands natural language, is also providing oncologists at the Memorial Sloan Kettering Cancer Center with improved access to current and comprehensive cancer data and practices. The resulting decision support tool will help doctors everywhere create individualized cancer diagnostic and treatment recommendations for their patients, based on current evidence – and has the potential to include available SNP research.
Watson is already being used by other healthcare providers. WellPoint, the largest health benefits company in the US is using the technology to provide alternative options to proposed treatment processes. "Just think of what Watson can do for physicians when it comes to answering difficult questions by looking up and cross-checking information, and providing a probability of success that this is the right answer," said Rick Kaplan, newly appointed Country General Manager of IBM Israel.
Laws and regulation could make new data available
In a panel discussion on how the availability of real world evidence could influence medical practice, Dr. Nava Sigelmann-Danieli Director of Oncology Service line at the Maccabi Health Services, pointed out that the individuals participating in clinical trials don’t accurately represent real world patients. For example, most women participating in clinical trials for breast cancer treatment are between the ages of 40 and 60. But in reality, most women suffering from breast cancer are over 70 years of age.
Panelist Dr. Lior Soussan-GutmanManaging Director of Oncotest-TEVA business unit in Teva Pharmaceuticals pointed out that new regulations requiring pharmaceutical companies to share the “real world evidence” collected during and after clinical trials would open another stream of data available to healthcare providers.
This combination of research, machine learning, and new laws continues to offer a more complete – and personal – view of healthcare treatment options.
Read more about the workshop presentations, here.
Machine learning and data mining at IBM Research – Haifa
The Machine Learning and Data Mining group specializes in developing algorithms that learn to recognize complex patterns within rich and massive data. Read more.
IBM's Biomedical Analytics Platform Helps Doctors Personalize Treatment
Italy's Istituto Nazionale dei Tumori testing new decision support solutions for cancer treatments. Read more.