AI in Healthcare for All

When it was first commercialized, the steam engine cost much more than other power sources available – until it didn’t. The engine, developed to pump water from flooded mines, allowed for deeper and less costly digging of coal. Then came faster transportation with cheaper shipping of more products and – as accessibility increased with efficiency – more people.

The full promise of a breakthrough isn’t in what it does initially. It’s in what it enables, eventually.

The fourth industrial revolution of digital, data and artificial intelligence (AI) will similarly unleash healthcare’s new future. Just as steam-powered technology once paved the way to help humans do more, faster, better and easier, AI-driven predictive tools, digital caregiving and new understandings of “health” and “care” will pave the way for better outcomes. Efficiency, which comes at a higher cost, initially, will lead to greater effectiveness through better healthcare prevention, diagnosis and treatment.

Eighty percent of doctors say AI in healthcare is useful. It is already there in exam rooms. It’s triaging hospital and emergency department traffic, analyzing patient risk scores and identifying potential new therapies by simulating chemistry with computers.

Having come into play as a “physician’s assistant” of sorts, digital and AI innovations are now positioned to do more as they earn trust across the healthcare ecosystem. While a lack of trust in AI remains a barrier to adoption, at ZS, our consortium of academics, clinical practitioners and industry leaders have settled on three principles to frame the requirements for trust in AI:

1. Responsibility: There are some problems AI should not solve, making its intent highly relevant. Similarly, irresponsible management of data and algorithms can unintentionally instill biases into analyzes, with damaging repercussions for real human beings.

2. Competence: Innovations must work – and the health ecosystem will need to come to terms with what defines an acceptable margin of error. The grace afforded to a human physician who makes a single mistake is not yet afforded to computer programs that recommend cancer treatments.

3. Transparency: Being upfront about the limitations of digital, data and AI in healthcare can help maintain trust in the face of imperfect competency.

Early adopters of digital, data and AI in healthcare have already fostered breakthroughs that set the stage for a transition from skepticism to a start at trust and a leap from efficiency to greater effectiveness.

Greek startup – and ZS partner – Intelligencia.ai uses AI to predict the probability of new compounds’ clinical and regulatory success. Ezra helps radiologists detect cancer lesions quicker and more accurately via AI. Whisper’s AI-driven hearing aid helps people distinguish human voices from background noise.

These innovations also blur the understanding of what “health data” can be: voice-based digital biomarkers are a bellwether of a coming revolution in biomedical research and healthcare. Voice quality – a sign of patient health – combined with technologies that enable its capture on everyday consumer devices enables symptom detection and prediction with tremendous potential for research and development and clinical care.

With digital shifts and technologically empowered consumers and clinicians, “healthcare” itself can shift from an episodic experience to an ambient one, as the doctor is now both in the office and on the phone in someone’s pocket. Consumers largely welcome these changes – our survey of 4,000 adults in the United States showed that 73% want greater access to care anywhere.

And with healthcare being everywhere, all the time, we start to see efficiencies through practice. For example, in the United States, Intermountain Healthcare saved tens of millions of dollars over a few years by using natural language processing technology to collect patient surgery data.

Digital health innovation in and of itself is many things, one frequently being expensive. But cost-effective innovation at scale is transformational.

COVID-19 necessitated the accelerated scaling of digital and AI to broader populations. Ireland, for example, pioneered remote monitoring for patients with underlying respiratory illnesses to track their health virtually alongside health service staff while remaining in quarantine.

Acceleration is also happening within the pharmaceutical industry through AI-driven automation of therapeutic compound screening. Rather than humans conducting a few hundred or a thousand lab assays to discover potential new medicines, researchers are conducting millions by simulating chemistry with computers, identifying more compounds that could pass the regulatory process.

As AI in healthcare is advancing, the healthcare ecosystem is preparing for this fourth industrial revolution. Over half (56%) of life sciences leaders we surveyed say their company has the right management support to introduce more AI into their work, yet, 46% recognize they have a shortage of people with the skills to implement AI. The number of hospitals in the United States that have implemented AI has tripled since 2020.

Preparation, pace and promises are driving investment in digital and AI technologies, which accelerates their scaling up.

As healthcare’s fourth industrial revolution accelerates, we will get better at “getting better.”

If efficiency means the cost of a cancer diagnostic test can drop from $1,000 to $10, more people can use it – and use it earlier. In healthcare, earlier is a matter of life and death: Melanoma, caught locally, has a 99% five-year survival rate. That rate drops to 30% if cancer has spread through distant parts of the body.

In clinical development, the discovery of millions of new compounds means medications that are more likely to be effective and pass the regulatory process will be discovered. Using AI for a better probability of success will vastly improve the 90% failure rate from the new chemical entity to Phase 1 clinical trial with the Food and Drug Administration, a United States agency.

At a cost-effective scale, innovations such as digital advances, AI and personalized genomics can be a primary course of action, not an expensive last resort.

Efficiency, then, leads to more effective healthcare. Just as we leapt into a more advanced global society simply because inventors found a better way to pump water out of a mine in England, improving efficiency and reducing costs will allow for more effective medicines to come to market and reach patients who need them.