AI in Healthcare: From Promise to Practice

By: Saurabh Johri, Chief Scientific Officer, Babylon

We are in the midst of the fourth industrial revolution, where Artificial Intelligence (AI) is transforming every aspect of society. As we have seen with the release of the next generation of AI models and algorithms such as DALL.E 2, ChatGPT and AlphaFold, this technology accelerates productivity, automation and innovation wherever it is deployed. Experts have put AI into practice in fields as diverse as journalism and the arts, where AI is generating novel content, to the hard sciences, where AI is solving grand challenges and previously intractable problems. Yet despite its promise in perhaps the most critical aspect of our lives – healthcare – the integration of AI into routine clinical practice has remained elusive.

Why is this? AI presents unique technical challenges when applied in a healthcare setting, from data privacy and security to data management and interoperability. The healthcare system itself is also deeply complex. The pressure on clinicians to do more with less is a real issue working against the successful implementation of broad-reaching AI technology in healthcare. For true buy-in and engagement with the technology, providers need to see both the near- and long-term impact of their tech investment. This means that there need to be clear operational efficiencies driven by automation, as well as improved health outcomes driven by personalized and predictive models.

To solve these challenges, technologists and clinicians need to work together to co-design healthcare technology and products that reflect a deep, holistic understanding of the needs of patients, providers and health systems. My years of experience working with technology in healthcare has shown that clinicians’ involvement is a key part of the equation and a necessary part of the solution. 

Three Promises of AI in Healthcare 

For providers to initiate this paradigm shift and reap the benefits of AI in healthcare, the industry has to understand and integrate three key impacts of AI on healthcare: prediction, personalization and automation.

The predictive capabilities of AI will evolve the current model of reactive sick-care to one that is more proactive. With this proactive model, care teams will be better prepared to intervene early, thereby improving care management and outcomes as well as reducing costs due to complications.

AI has the potential to greatly improve the personalization of medical care by analyzing large volumes of disparate data, from sources like claims, EHRs, devices, and public health studies, and continually learning from them. By identifying patterns and risk factors, AI can help to develop personalized, real-time treatment plans for patients. The impact of AI on personalization holds the promise that patients will become more invested in improving their healthcare outcomes if they receive the right information at the right time. 

With respect to automation, AI has the ability to free up scarce clinical resources by automating basic clinical tasks and improving standardization and care quality so that care teams can focus on what matters most. My team at Babylon recently published peer-reviewed research demonstrating the impact of conversational AI technology in supporting the automation of clinical note summarization, which reduced administrative burden and delivered more focused clinician attention during consultations. 

The benefits of delivering predictive, personalized healthcare and automated workflows through AI-powered tools are clear. Yet many clinicians remain skeptical about AI, specifically the accuracy of the deep insights the technology can yield, its trustworthiness and vulnerability due to bias. Fortunately, new approaches in the field of Fair and Explainable AI (XAI) and ‘human-in-the-loop’ systems provide attractive solutions for alleviating these concerns. And industry leaders are already working on the next generation of intuitive, conversational technology products that will provide quick and easy access to AI within operational and clinical workflows.

Transforming Promise into Practice

How can our industry translate the promise of AI into clinical practice, in order to deliver value across health systems? One way to envisage the future of AI in healthcare is by taking a step back and using a non-clinical lens. Industry leaders can apply to the healthcare industry the same data, AI and engineering principles that have transformed the logistics, transportation, and travel industries.

Modern cloud-based data solutions, for example, can provide safe, secure and scalable systems for ingesting and structuring large volumes of diverse healthcare data – the fuel for AI models and algorithms – to deliver personalized and predictive insights that improve patient outcomes. Continuous monitoring through AI models can assess changes in patient status in real time and deliver recommendations as needed. 

Additionally, modern product design principles which skillfully integrate intuitive conversational AI experiences can help deliver the power of AI-derived insights to patients and clinicians when they need it most.

Uniting prediction, personalization, and automation – the promises of AI in healthcare – technologies like Generative AI, Natural Language Processing and AI-enabled clinical protocols are helping clinicians more efficiently deliver counsel and communication to their patients. By continually learning and providing more targeted interventions, we expect to see increased patient engagement and associated improvements in health outcomes. By fully embracing these technologies, the wider healthcare industry can shift from a reactive model of “sick care” to a proactive model where prevention is prioritized.

This is the ultimate promise of AI in healthcare. With AI and clinicians working together to improve patient health outcomes, if we can predict it, we can prevent it.