Today’s life sciences and healthcare continuum is fully enhanced by an ever more connected array of sensors and machines. In this setting, pharmaceuticals and medical devices are developed and manufactured in facilities that are fully monitored by sensors. Devices used by clinicians, researchers and patients alike are monitored by an increasing number of sensors. Medication adherence will also be further managed and monitored by sensors. In aggregate, the capture of data from patients provides real-world data of use, misuse and treatment response, on an individual basis.
The randomized clinical trial (RCT) has been the gold standard for the efficacy of medical interventions. RCTs, however, face two main obstacles: recruitment and retention. Recruitment requires identifying patients who meet a set of specific criteria, which is difficult and time-consuming.
Between a quarter and two-thirds of trials fail to recruit their required number of participants. Access to larger pools of patient-generated health data (PGHD) — much of it derived from connected sensors and health and wellness apps — enables the identification and recruitment of suitable study candidates.
Once patients participate, data from wearables and other monitors provide closer contact with the study team, while reducing the need for the subject to travel to a research center for periodic evaluation, which improves retention.
Such monitoring also provides more precise information about adherence to the study protocol. Efficacy numbers can then reflect the actual use of a medication, as well as reveal the full impact of side effects or other obstacles to proper use, providing novel insights to effectiveness. Engaging patients in managing their own care in trials may also streamline future clinical development workflows.
Real-world data (RWD), gathered from regular monitoring of a wide range of patients over long periods of time, provides detailed information about how treatments work in daily life for more patient types than could be included in an RCT and, once analyzed, becomes real-world evidence (RWE) that enhances therapeutic effectiveness. RWE is also used to monitor post-market safety and adverse events for both biopharmaceuticals and medical devices, improving care decisions and therapeutic guidelines.
RCTs can fail to provide meaningful insight when some part of the population responds strongly to a treatment while others don’t. RWE can help stratify patients into subpopulations that are more susceptible to a particular disease or have a specific treatment response. Additional data, such as genomics and other biomarkers, may help identify the mechanisms behind these responses. Combined, these expanding sets of clinical and ambient data allow for more specific targeting of patients for the appropriate trials.
Many diseases are increasingly seen as a complex of several diseases with similar symptoms, which can be divided up and defined by mechanistic pathway (endotype), specific clinical presentation (phenotype), and individual patient susceptibility (genotype).
This growing knowledge of diseases can enable a more precise balancing of interventional benefits and risks. For example, Herceptin is a breast cancer drug that is an effective treatment for women with too many Her2 protein receptors in their tumors, which are caused by an overexpression of the Her2/neu gene. Herceptin works less well for tumors without excess Her2 receptors, but for all patients it increases the risk of heart dysfunction. So, a test for the Her2/neu gene enables targeting this treatment for maximum benefit while minimizing exposure of some patients to unnecessary risk.
A more granular understanding of disease states leads to better-defined RCTs and more narrowly defined and targeted treatments. Such precise therapies will increasingly challenge manufacturing processes as they demand low-volume production runs of specialized therapies aimed at specific populations at the same time as overall capacity is kept high.
Even narrower are personalized therapies that use the patient’s own cells as the basis for treatment. An example is autologous CAR-T cancer therapy, which extracts T cells from a patient’s blood, genetically reengineers them to recognize and kill cancer cells, and then infuses those cells back into the patient. Currently, these processes can’t be carried out in the same location and require a complex supply chain with strict temperature and custody regulation and tight timelines.
Looking forward, facilities may grow smaller and more localized, promising more localized manufacturing, shorter supply chains and quicker response.
Sensors and intelligent machines are also transforming biopharmaceutical manufacturing by providing visibility into the process rather than just the equipment. Connectivity to devices by industrial control systems has been around for a while but was difficult to modify and required the attention of a large, skilled staff. This new level of connectivity and automation makes it possible to improve manufacturing system efficiency, quality and reliability, increasing yield capacity and driving revenue.
For example, previously, measuring cell density in a bioreactor required taking a sample to quality control and waiting for the result before deciding to add more glucose substrate. A closed-loop control using a Raman spectral probe to continuously measure cell density enables the precise addition of glucose multiple times daily to maintain the optimum density and maximize yield.
Currently, there can be significant quality and yield variance from one facility to another and from one batch to another in the same facility. These variances stem from a number of factors including employee procedures, equipment maintenance, raw material quality, facility environment and other causes. Connected sensors that measure every parameter can finally provide a view into every process step and its downstream consequences.
For example, recording the exact time that a bioreactor is inoculated with a culture and then being able to tie that time to downstream results, combined with knowledge of all other varying inputs, provides a previously unavailable context and enables understanding of the root causes of variance in quality and yield. Those causes can then be addressed to minimize variance.
This real-time closed-loop control data has a positive effect on R&D as well, as information about manufacturing constraints affects initial pharmaceutical research and design. For example, mammalian cells are susceptible to shear. Various processes in a bioreactor, such as agitator ramp-up speed, gas sparging to provide oxygen, and foaming that leads to bubbles bursting, can all result in shear. These parameters of the manufacturing process, along with many others, need to be taken into account to design maximum productivity into a therapy from the beginning. This links drug development to the GMP manufacturing facility, and also generates the data that aids in regulatory compliance.
This real-time and contextual manufacturing data enables changes up and down the supply chain, optimizing end-to-end throughput, and meeting the increasing challenge of smaller batches required by precision medicine.
As indicated in our initial installment of this series, the information flowing up from individuals finally returns to them in the form of more effective and easily used treatments, enabling patients to benefit from treatments that are effective, timely and personal. The interconnected world of sensors and machines is transforming all aspects of the healthcare and life sciences industry.
This article was written by Rodan Zadeh, Head of Connected Care Strategy within Cognizant’s Digital Business, and Brian Williams, Chief Data Officer within Cognizant’s Life Sciences Practice.
Also contributing to this article were David Staunton, Sashi Padarthy and Naveen Nayar from Cognizant’s Life Sciences, Healthcare Consulting and IoT practices.