Segmented and Personalized Medicine using Data-led Insights | HCL Blogs

Segmented and Personalized Medicine using Data-led Insights

Segmented and Personalized Medicine using Data-led Insights
July 14, 2020

Co-authored by : Dwaipayan Chakraborty

Did it ever occur to you that the same drug, when administered to you, did not cure the ailment as much as it did for others you had heard from? Or, the side effects, you had from the last pill prescribed to cure your headache or joint pain, were the cause of greater misery than the headache or pain itself. That’s because individuals have a very different response to a drug based on their genetic and metabolic profile. This is particularly true when the drug itself only targets a specific genetic subsection of the patient.

To target this issue a specialized field known as Individualized or Precision or Personalized Medicine, a branch of Pharmacogenomics, has been exciting the medical community for a while. These terms describe a genomically customized medical regimen for individual patients. These therapies promise treatment through exact dosages and accurate medicament specifically designed for an individual with a set of genomes. The focus is on developing digital technology-enabled processes to create tissue grafts with improved in vivo regeneration incorporated into the current tissue engineering protocols.

While, the financial viability of such a therapeutic model may scare away most investors, digital technologies can make, population-based segmented approach to medicine, a real possibility in managed markets. So, although the medicine will not be entirely personalized, it will target a sub-segment of the population through a targeted gene pool.

As healthcare advances more and more to a managed market setup, payors are incorporating new methods to compensate drugs that look for higher efficacy and lower side-effects. However, such efficacy and outcome studies are carried out post-facto and the results of one population may be significantly different from another. Today payors are exerting tremendous pressure and using outcome-based models to cap drug prices thereby limiting the freedom of pharma companies to price their products. To respond to this evolving dynamic where managed markets demand greater patient outcomes, the aim of pharma companies should be to exclude drugs from a market where it is ineffective and charge a premium to those where it is more likely to work.

The last decade saw a tremendous investment in digital technologies by pharma companies in being more patient-centric across their commercial and R&D operations. This has created a more personalized treatment approach in care models. With the influx of personalized data analysis from these engagements, pharma companies will have a treasure trove of information that can be harnessed for biomedical research in segmented medicine.

Although, these patient centric models have tried to create a single patient story, the data mining at each phase of this story is stored in different silos. Running prescriptive and predictive analytics on top of this data analysis would require capabilities to derive insights by integrating disconnected datasets. AI/ML based data mining capabilities can help to provide domain feeds based on the principles of genomics, metabolomics, and transcriptomics to build a prediction model. This can help to overcome the hurdle that arises with multi-domain integration.

The advancement in cloud computing has taken away the burden for large-scale on-premise infrastructure that biological computing demanded, and big data analytics is no longer a cutting-edge tool but features as a necessary capability in the assessment of disease, patient, and molecular profile. A combination of mobile technologies, cloud, and IoT will need to work in tandem to accelerate the pace of data analysis and extrapolate the results.

The initiation steps will see repurposing existing drugs for more targeted usages in certain patient segments. Precision and segmented medicines need to gain insights into existing drugs and their efficacy in other areas. Big data analytics using predictive and prescriptive tools can help foster an in-silico simulated experimentation, which will accelerate and augment the existing in-vivo and in-vitro pre-clinical procedures.

Figure 1: Methods of analyzing data for patient profiles and patient clusters
patient profiles and patient clusters

Once the data structures for patient profiles are presented in a consumable way to the data scientists, the real research will begin and shall demand greater prowess to design analytics models for studying the massive volume of aggregated biomedical data, medical images, and videos. This needs to leverage a platform that has deep-learning capabilities along with functionalities to observe, interpret, evaluate, and act on the insights. The biggest challenge will be to design application and IT infrastructure that can deal with the data mining abundance and complexity of biological systems.

Tissue engineers should adopt machine learning algorithms and high-end predictive computing models for gene-identification, manipulation, and editing. The same should be used to create cells to constructs personalized tissues to develop personalized medicine.

Where are we today?

FDA has granted approval to many medicines that were non-efficacious in a larger sample size but has shown proven efficacy on individuals with a genotype. Well-designed projects by pharma companies in this arena have forced regulators to streamline and expedite their internal review procedures. Several cancer drugs are the known for such kind of research, for example– crizotinib, vemurafinib, dabrafenib, and trametinib

Pharma companies are making significant investments in digital technologies to become increasingly patient-centric

This has a profound implication on the digital strategy of a pharma company. Segmented medicines will require advance analytics and intelligent insights for successful operations and the in-vitro in-vivo model of testing that exists today will need to indispensably add an in-silico component. Besides the analytics required for drug development and trials, the niche markets that these segmented medicines would create will require companies to focus on better market analytics and more experimentation and innovation in their interaction with customer groups.

Clinical Research

This demands for reinforcement of informatics-oriented study design with primary and secondary endpoints for disease treatment or prevention that takes care of individual variability around genetics, ethnicity, or lifestyle. Protocol design should not only be genetics-centered treatment but should be more dynamic based on risk assessment facilitated by high end computing for treatment optimization. Diseases with unknown or of less understood etiology form one of the ideal candidates for this type of research. Prediction modeling strategy should be leveraged to recognize disease markers.

This could also result in more AEs (adverse events) when the drug is not delivered to the correct segment. To avoid this scenario the risk for hypersensitivity to the investigational product can be calculated based on a genetic test. Social media listening combined with thousands of computerized simulations (based on electronic health records (EHR)/personal health records (PHR) etc.) can help to reach out to the targeted subjects. These subjects can be pursued to participate in the trials after confirming their genetic profile based on predefined study criteria. Since sample size for such studies are usually very less, with a non-homogeneous phenotype, so it would be of significant impact even if a single patient tends to drop out, thus patient retention will also need considerable effort. CDM (clinical data management) tasks have to be done with computer-intensive data processing prior to its analysis. Interdisciplinary cooperation of subject matter experts is the key to an outcome-based approach when it comes to this kind of data analysis.

Overall Approach

Adaption Roadmap
Figure 2: Adaption Roadmap for Pilots in Segmented Medicine.

The vital aspect of personalized medicine is to provide patent-centric care. To make personalized treatment a reality, patient populations and segments need to be identified better through genomics, proteomics, and epigenetics. The progress to this state will be in phases and shall require collaboration between multiple industries leading to a new therapeutic revolution

Science needs the support of technological advancement both for biologics as well for small molecules so that ‘targeted therapy’ combined with ‘targeted diagnostic tests’ for ‘targeted gene set’ can provide paradigm shift from conventional medicine to a contemporary personalized medication which is harmless and affordable.