Big Data And Virtualization In Precision Medicine

As science continues to advance at unprecedented rates, new insights are being provided into the biology of cancer that prove disease is an expression of patient-specific genetic characteristics. In cancer, mutations - a change in DNA sequence, similar to a misspelling - occur in important genes, resulting in uncontrolled cell growth and proliferation. Each patient's cancer has its own unique set of mutations. The current one-size-fits-all treatment paradigm fails to account for these unique set of mutations, resulting in low response rates. Precious time is lost while physicians progress through successive standard therapies with no prior guarantee of efficacy.
 
Precision medicine seeks to objectively analyze unique disease characteristics, such as genetics, to personalize therapeutic treatment for each patient's set of mutations. It demands the reinvention of the design, testing, and prescription of treatments. Specifically, treatment design should be more effectively targeted, testing should be focused on likely responders, and prescription should be based on a match with an individual patient. By doing so, precision medicine offers the promise of averting unnecessary treatment, minimizing drug adverse events, and maximizing overall safety to ultimately maximize the efficacy and efficiency of the healthcare system. The ability to rapidly identify the most beneficial therapy for each patient will transform the healthcare experience.
 
Today's technology is ushering in the age of Big Data. 'Big Data' - characterized by large volume, velocity, variety, and veracity of data - in healthcare is growing exponentially. Genetic sequencing technologies, such as next generation sequencing, have made the analysis of each patient's mutations accessible, feasible, and increasingly commonplace. Non-invasive monitoring approaches enable the collection of patient data longitudinally across multiple time points. Additionally, there is an ongoing availability of new information on disease biology, with over 3 million publications available in PubMed for oncology alone.
 
To fully realize the potential of precision medicine, we must first "unlock" health Big Data. The inherent nature of Big Data presents many challenges in this endeavor. The average cancer contains hundreds of interrelated and dynamic mutations. It is impossible for the human brain to manually map and understand the implications of ongoing interactions between genes, multiple layers of biology, and drugs. Thus, we restrict our analysis to a mere scattering of available data points. Precision medicine use of genomic data, for example, is particularly constrained, utilizing a selection of mutations that represents less than a trillionth of a percent of the entire human genome. By forcing reliance upon correlative rather than causal relationships that may be spurious or only account for negligible disease impact, the validity of genomic studies is compromised. Ultimately, the challenges of Big Data preclude the visualization of clear and concise clinical action that provides value to the patient, physician, and healthcare system.

In the face of the many challenges of translating Big Data to personalized clinical action, cognitive systems capable of sensing, predicting, inferring, recommending, hypothesizing, and reasoning are needed. For precision medicine, we need computers systems capable of incorporating the large volume and variety of Big Data from scientific publications, omics databases, multiple levels of individual genomic data, drug data, and many other potential data sources, into virtual simulations of patients, diseases, and drugs. These virtual simulations can then be utilized to study cause and effect of individual biological changes, biological interactions, disease progression, treatment mechanisms and efficacy, and even simulate drug trials or discover novel uses of existing drug therapies.
 
Precision medicine is already offering great hope in cancer. In September, Keytruda, was released in India for treatment of advanced melanoma. An American blockbuster drug made famous by former President Jimmy Carter, Keytruda is also used in the United States for the more prevalent non-small cell lung cancer (NSCLC), as well as Hodgkin lymphoma, urothelial carcinoma, and head and neck squamous cell carcinoma. Unlike standard cancer treatments which are given to large populations of patients, Keytruda is a precision drug targeted to cancer patients with two specific biomarkers, making it the first drug to be approved based on the presence of molecular biomarkers rather than tumor location. A 21-day supply of Keytruda is estimated to have a maximum retail price of 2.36 lakhs, or nearly 0.5 crore a year. Given that the identified biomarker signature represents only 4% of cancers and only 30% on average of these patients respond to Keytruda, a more reliable means of predicting response is still needed to minimize cost and maximize efficiency. Cognitive systems show great promise in this undertaking. Using virtualization technology, an additional 24 potential biomarkers have been identified. These biomarkers can be used to more create more robust selection criteria that accurately predict and stratifies treatment response, as well as can be used in the development of combination therapies capable of increasing response rates.
 
Going forward in the race against time, healthcare must focus on utilizing cognitive computer systems to combine Big Data with virtualization. The healthcare community urgently needs reliable means to match the right therapy to the right patient. All the tools are available, we must simply discover increasingly efficient ways to reach valid, actionable conclusions benefiting all healthcare players.