Weighing the pros and cons of the on-premise vs cloud data analytics debate in the context of health data, Raj Srinivas, CTO, 8K Miles Software Services summarises that if we bring overall cost of ownership and future ROI into the picture, the value-addition offered by cloud analytics solutions can be multi-fold. Additionally, with the recent advancements in cloud analytics, meaningful insights into vaccine effectiveness for COVID-19 mutations on individuals can be gained. Eventually this should result in providing personalised and quality healthcare solutions to individuals, throughout the world
As COVID-19 vaccine administration has commenced in parts of the world, individual and personalised healthcare will be the prime topic of 2021 world-over. Even though the trial results of various vaccines are promising, any adverse performance of these vaccines even on less than one percent of world-wide population (given the fact that we have a more mutated strain of COVID-19 now) could affect hundreds of millions of people. Given this situation, it may be prudent to have post-vaccine administration scenarios pre-analysed from patient history data so that, proper precautionary measures can be adopted prior to the administration of a vaccine to an individual. Attributes affecting the vaccine effectiveness on individuals, such as current usage of medication, allergies, hereditary traits, co-morbidities and any other medical condition of the patient all need to be studied thoroughly, as a larger world population gradually gets vaccinated.
The challenges post COVID-19
So how do enterprises go about analysing years of medical data of billions of patients and compare them with COVID-19 clinical trial data to arrive at decision data models? Patient data from EHRs, digital data from hospitals/clinics/labs, imaging and genomics data are some common data sources to start with. Ingesting, storing and analysing this data in on-premise data centers is an option and so is a cloud-based data analytics platform. Patient data is going to grow enormously and soon pharma companies can be looking at petabytes and zettabytes of data.
In order to store, clean, govern and analyse data of this scale, constant need for additional computing resources that can be provisioned/de-provisioned in a Just-in-time (JIT) fashion, is a dire need. Additionally, we are looking at data scientists of pharma companies logging on from different parts of the world, continually consolidating ever-changing data to perform nimble vaccine predictions and trend analysis on a 24×7 basis, in a water-tight data access security, governance and data audited environment.
On-premise vs cloud data analytics
Will the on-premise data analytics model be able to stand and deliver given this set of harsh requirements or is cloud data analytics platform a better option? Let us investigate this.
- Cloud data analytics platforms can auto-scale to accommodate huge volumes of patient and clinical trials data, and their analysis module runtimes, on-demand, by vertical and horizontal scaling/de-scaling of instances, memory, cache, storage space and CPU resulting in flexible/low costs. Whereas in an on-premise setting all of these resources have to be procured upfront and is non-elastic over the course of the analysis phase.
- Most of the cloud data analytics cost come in during the time of patient data analysis iterations and storage, and not during the provisioning of the data analytics infrastructure in the cloud. But in an on-premise model we begin to incur cost from the time we provision hardware.
- Patient data access to the right kind of users/roles can be predefined and effected by way of policy in the cloud data analytics platforms resulting in granular access control regarding who has access to data and audit capabilities to understand who accessed the data. Some cloud data analytics platforms mandate multi-factor authentication mechanisms and PII (Personal Identifying Information) restrictions especially for patient PII data-sensitive data lakes. However, in on-premise situations this may be a cumbersome process to implement for a global set of data analysts dealing with structured and unstructured patient data that need to be sourced from different end points.
- Patient and clinical trial data stored and analysed in cloud allows the data analysts to access and work on the analysis from any computer in the world even if the on-premise access or VPN access is absent. In on-premise situations, VPN or such secure access is a must for global and local users. If such access fails, it leads to a single point of failure.
Potential of cloud data analytics to yield meaningful insights
In the post COVID-19 world, we see that enterprises in the healthcare and pharma industry, are moving their on-premise data and clinical trial data to cloud data analytics platforms that are endowed with proper built-in dataOps, data availability/accessibility modules and with augmented data management processes. If we bring overall cost of ownership and future ROI into the picture, the value-addition offered by cloud analytics solutions can be multi-fold.
Gartner in October of this year has predicted that by 2022, public cloud services will be essential for 90 per cent of data and analytics innovation. In addition, by the end of 2024, 75 per cent of enterprises will shift from piloting to operationalising AI, driving a 5X increase in streaming data and analytics infrastructures in the cloud. As a top 10 trend in data analytics for 2021, Gartner has said that “cloud is a given need” for data analytics this year.
With the recent advancements in cloud analytics, meaningful insights into vaccine effectiveness for COVID-19 mutations on individuals can be gained. Eventually this should result in providing personalised and quality healthcare solutions to individuals, throughout the world.