How artificial intelligence and machine learning are transforming the healthcare industry

Access to basic healthcare is seen as a fundamental global right. However, in both developed and developing economies, we find the surge in chronic lifestyle ailments and rising populations overburdens healthcare systems. The pandemic exacerbated the situation, stretching the patience and resilience of healthcare workers to the maximum as well as highlighting the economic divide between the haves and have-nots.

When we refer to access to healthcare, we often restrict our thoughts to the ability to provide timely medicines to underprivileged people globally or diagnose ailments in time. Other forms of access are disguised and, for years, still need to be addressed. Treatment of rare diseases is a problem due to the long cycle and unfavorable economics of drug discovery. Many patients are prescribed generic cocktails of drugs as a part of their therapy which has multiple side effects; personalized medication is infeasible. Moreover, our healthcare infrastructure is not equipped to handle spikes in health situations due to a lack of planning and resources. These problems are silently costing patients their lives.

Against this worrying backdrop, the emergence of artificial intelligence (AI) and machine learning (ML) has provided a ray of hope to both physicians and patients. By harnessing AI-enabled tools, workflows and procedures are streamlined, making healthcare more affordable, personalized, effective, and equitable. With AI and ML, healthcare providers are better positioned to diagnose, treat, and prevent ailments.

Increasing Adoption and Advantages of AI/ML

We are in the midst of a silent revolution driven by algorithmic technologies disrupting the healthcare industry. More healthcare organizations are gradually using some form of AI in their regular operations.

Healthcare has always been data-rich with sources like electronic medical and health records, claims, genomics, and clinical trial data. Advances in technology have now allowed us to integrate these data sources cost-effectively and timely while maintaining data privacy laws. The development has opened up many new business-use cases where we can apply AI/ML techniques to find patterns that humans would fail to detect and generate insights that were not feasible earlier.

Let us take an example of AI-generated drugs. Traditionally, the life cycle of clinical trials would take 8 to 10 years on average. Of these, many drugs would not get FDA approval or fail clinical trials, making it cost-ineffective for Pharma companies to invest in rare diseases with a limited market.

But AI has changed this. Key time-taking activities in R&D, such as disease modelling, target discovery, and molecule identification, are now algorithmically driven using deep learning and NLP techniques. Empirically, this has improved the success rate of clinical trials and reduced the time to market life-saving drugs, making it cost-effective for pharma companies to venture into rare diseases. There has been a spurt of many “As-a-service” AI start-ups specializing in these activities, which are partnering or being acquired by big pharma. Examples are AstraZeneca’s collaboration with BenevolentAI and Sanofi’s collaboration with Insilico Medicine . In 2022, as many as 18 AI-created drugs were in the clinical phase, a mere reflection of what the future will look like.

Another application is the personalization of treatment. It is seen that every individual’s reaction to treatment is biologically different based on genetic, behavioural, and environmental factors. With the maturity of genomics data and its combination with diagnostics, medical records, and Health monitoring data, an opportunity has arisen to create personalized or precision medicine. Personalized therapies are being tailored for individuals using techniques like deep learning and neural network algorithms. Therapy areas such as Oncology and immunology as early adopters in this area.

Diagnostics is another area with the immediate relevance of using AI. Self-learning algorithms can be trained to identify, from images and sounds, any anomalies within the diagnostic reports. These augment the medical practitioners’ diagnosis and help overcome situations of practitioners’ capability, availability, and fatigue. Now imagine if these algorithms were embedded within devices and made accessible to the underprivileged and marginalized strata of society – giving them access to patient outcomes and early diagnosis of ailments that they would usually not be able to afford.

There are many more applications of AI in Healthcare; however, the one that I would like to highlight deals with epidemic modelling and prediction. Recently we saw the world brought to its knees during the COVID pandemic. We must not be under any illusion that this was the last time such an occurrence happened. How can we learn from our past and simulate responses to mitigate any such future pandemics? One way is to create multiple simulations using AI-based forecasting techniques on how a pandemic can originate and travel globally. And design responses to each simulation involving collaboration between governments, healthcare ecosystems, and global citizens, thus saving lives in the future.

Addressing Privacy and Data Quality Concerns

Despite these advantages, there are apprehensions that using AI-enabled tools in healthcare can increase ethical concerns. For instance, if AI/ML algorithms are not programmed appropriately, there is a threat that human prejudices could be perpetuated through these tools. Nevertheless, judicial and impartial human oversight can ensure such biases do not flourish.

Thus, the quality of data used as inputs to these algorithms is essential. Secure data lakes need robust data governance and quality regimes to build confidence in the data before handing it over to the algorithms. We tend to trivialize this area, but it can destroy any algorithm-based ecosystem.

Considering the burgeoning amount of sensitive information received daily by healthcare workers, privacy measures and data ethics remain paramount. Since patients willingly trust healthcare professionals by disclosing personal information, clinicians are responsible for ascertaining that this data is always safeguarded zealously.

Data safeguarding includes:

• Using secure storage systems
• Limiting data access to authorized personnel only
• Procuring informed consent for using any patient information

Therefore, as technology keeps advancing continuously, healthcare firms must stay up to speed vis-à-vis the latest data privacy rules and software that maintain patient trust by safeguarding sensitive information.

In conclusion

These are exciting times for the Healthcare industry, where algorithm-driven economies drive disruption. Insights are generating favorable patient outcomes and moving the world towards greater healthcare access. Hopefully, this will lead us to our dream of democratizing healthcare for all.

Author: Rohan Nag, Senior Director, R&D, Axtria