Analytics and Big Data in Pharma
Posted 17th September 2018 by Kieran Chambers
Drug discovery and development is a complex process that requires integration of multiple data points, experiments and calculated risk/benefit assumptions. It is therefore only natural that virtualization and big data analytics are a natural fit for implementations, expected to demonstrate significant cost-effective gains for the betterment of society, providing access to effective and safe medications.
We all know that technology and big data analytics are rapidly evolving to revolutionize virtually every aspect of our lives. Most parameters of relevance these days can be measured by wearables, implantables, and ingestibles. Technology, big data analytics and artificial intelligence are now demonstrated useful in health management and drug discovery and development. Patients and physicians progressively adopt the technology. Pharma is the next frontier of the digital revolution.
Digital Clinical Trials
The conduct of traditional clinical trials is relatively ineffective, incurring high costs and lacking productivity justifications. The transformation of clinical development process is contingent on pharma’s ability to execute patient-centric smart and virtualized trials.
- First and foremost: patient centricity. Sensors allow us to conduct clinical trials with significant fewer burdens to the patient. Most of the trial can be done remotely at the patient’s home and natural environment/routine. The reduction in number of trips a patient needs to make to the clinic/hospital, missing a day of work, troubling family members translates into better experiences with health management and better compliance to therapeutics.
- Virtualization is a means to objectively fill the data gap – over 98% of patient data is in fact missed and untracked during the conduct of regular trials, since the data is often collected at physician visits only. By utilizing sensors and continuously collecting data, pharmaceutical companies can eliminate this gap. Real-time, continuous data allows cancelling out food and mood effects, as well as other fluctuations, which render the data more robust.
- Using sensors and algorithms can potentially mitigate errors caused by the human factor of the medical assessment – these may include inconsistencies amongst physicians and subjective non- reliable self-reports of patients. These improvements can allow for more objective monitoring of disease progression, better measurements of clinical endpoints and help design more efficient drugs and therapies.
- Using advanced analytics can reveal the influencing factors on the dropout rate in clinical studies and contribute to increase in patient retention. Better site selection and smarter ways for recruitment can result in enrolling fewer patients, thus reducing study timelines and budget.
Advanced Analytics in Disease Management
The day-to-day management of disease is burdened by the active investment required to fill out questionnaires, take multiple prescription pills and monitor safety and progression parameters. Passive wearable and big data analytics technologies are ripe to provide effective alternatives at the comfort of people’s homes.
- The accepted clinical measurement of disease symptom severity is often based on subjective and potentially recall-distorted reports, as well as variable and semi objective observations by the clinician. The lack of measure continuity and objectivity generates biases in diagnosis, progression and state of disease, and adversely affects the prescribed treatment plan.
- Long-term monitoring of disease parameters allows a system to detect and flag changes requiring intervention, thus improving the care patients are able to receive.
- Tools for continuous and objective monitoring can establish digital markers of disease progression.
- Application usage is expected to increase treatment adherence
To summarize, by applying machine learning to analyze vast amounts of data from different sources, computers can help us identify complex patterns and the slightest changes within them, without being explicitly told where to look for the relevant data. Instead of relying only on traditional predictors, such as family history and few blood lab parameters, machine learning evaluates dozens/hundreds of factors simultaneously to help physicians and drug developers identify risk factors and response predictors.
Pharmaceutical companies should aim to use analytics and big data to improve the health and wellness of people everywhere by pioneering the implementation of digitally enabled medicines, catering to people’s health needs throughout their lives.
Lena Granovsky is the Associate Director of Analytics and Big Data at Teva Pharmaceuticals.