skip to content

Executive and professional education

 

We spoke to Janak Gunatilleke (EMBA 2013) about his new book, Artificial Intelligence in Healthcare: Unlocking its Potential, and his career journey.


Janak Gunatilleke.

Janak Gunatilleke is a qualified doctor who started his career in the NHS when he came to realise that medicine was too specialised for him and he wanted greater variety in his work. He moved away from medicine and into consultancy, while remaining in the healthcare sector. As his career progressed, he looked for more theoretical knowledge of the core business areas in which he was working and decided to return to school to study for an Executive MBA (EMBA) at Cambridge Judge Business School. Janak chose the EMBA as he was keen to study with experienced and like-minded people from a variety of different backgrounds. What attracted him to Cambridge Judge rather than other business schools was the local entrepreneurial ecosystem; this proved to be an inspiring environment in which to explore different technologies and the world of startups.

Janak’s interest in technologies continued post-EMBA and he became involved in data and AI, while remaining in healthcare. “Healthcare is the right area for me to be and I think that what I’m doing now around data and AI within healthcare is my sweet spot.” This focus ultimately led to him writing his recently published book, Artificial Intelligence in Healthcare: Unlocking its Potential. In this book, Janak takes a holistic approach to the design and development of AI solutions in healthcare to improve the likelihood of successful implementation and, later, scalability.

Successful implementation of AI in other industries

Janak explains that there are several barriers to rapid implementation of AI in healthcare and, to fully understand these, it’s useful to first look at areas in which AI is currently being implemented successfully. These include the commercial sector where AI is being used by companies such as Netflix, Uber and Amazon. These companies have access to huge datasets and use AI to solve a real problem, which adds value for the customer, whether it’s making a good recommendation on Netflix or Amazon, or improving warehouse efficiency, which reduces costs. “In these companies”, says Janak, “there is room for error; content recommendation on Netflix might not be perfect, but it’s not a question of life or death. In healthcare, the stakes are much higher; a wrong decision can have a negative impact on a patient’s life.”

AI in healthcare: challenges and ethics

Janak identifies three broad areas in which the use of AI presents challenges in healthcare; these are people, systems and technology. “People tend to lack trust and confidence in AI-based systems to augment human decision-making. In addition, there are also cultural barriers to adopting new technologies in general.” From a system perspective, the use of AI in healthcare is currently fragmented with isolated pockets of innovation. Each individual solution solves a specific problem or performs a particular task. “It isn’t easy to join the multitude of different systems seamlessly, resulting in difficulties in usage and a lack of interaction”, says Janak. This creates further barriers to adoption.

One crucial aspect in terms of technology is data. Healthcare data is personal and very different to the data available to a company such as Netflix, which has huge volumes of data that it has collected through its own systems. Patient data is subject to privacy laws and the appropriate safeguards must be in place to maintain patient confidentiality. Patient data exists in varying volumes within different systems and founders and innovators must overcome various obstacles to gain access to it.

The implementation of AI in healthcare also raises ethical issues. These include bias and generalisation which, according to Janak, “essentially means making the wrong decisions in the context of a particular group of people based, for example, on ethnicity or gender”. To give an example, an algorithm used in a study in the US to manage the healthcare of millions of Americans effectively discriminated against black patients because it used cost as a proxy for health needs. As less money is spent on black patients, the algorithm concluded that black patients are healthier than white patients with equally serious health issues.

Data cleaning and awareness can help prevent data bias and are essential to the quality of AI decisions. Janak takes Cambridge as an example; it’s a city that is made up of a certain mix of socio-demographic factors, which might not be representative of other areas. Janak explains that “AI is learning based on the data with which it is provided. It is ultimately only as good as the data and the people who use it; if a model built on non-representative data is applied in a slightly different context, it won’t perform as expected.”

Scaling the use of AI in healthcare

For AI to be scaled up successfully, Janak pinpoints three essential elements. The first is identifying a problem and then the technology that can be used to solve it. “A lot of the time, I see people getting really excited about some clever model or algorithm, and then they try to shoehorn it anywhere; it becomes a case of the solution trying to find the problem.” A solid case needs to be built for the right application, validated with stakeholders. “You’re not going to get the buy-in and the time and resources if it’s something that people think is very low on the radar,” says Janak.

Secondly, the right processes need to be developed and shared. “Whether you’re thinking about your idea, starting to gather data, developing the model or implementing your solution, there are different things that you need to do, depending on where you are on the journey and who you are.” For example, healthcare operators need to work with innovators and end users to support the development and refinement of the value proposition during the ‘design’ phase. “One of the key things is that nobody can do this by themselves; it’s important to bring all the diverse contributions together, and that everyone sees where they fit in the jigsaw.”

Finally, certain ecosystem or infrastructure elements must be in place, “I call them national enablers,” says Janak. These include access to good quality data and curated data sets. With the appropriate controls in place, certain data can be made available to innovators for building models. There is also a need for specific funding and evaluation models that capture the requirements of the AI entrepreneur. “None of this can be achieved in isolation, rather it has to happen at policy or national level.”

The future of AI in healthcare

In radiology, AI is currently being used successfully for analysing diagnostic imagery and optimising workflows, and there is huge potential for AI to add value and have a positive impact in many other areas of healthcare as well, for example, in what Janak describes as the ‘back-office’ areas, namely the operational rather than the clinical side of healthcare, where there is less risk and fewer restrictions in terms of getting products out to market. The data involved is less sensitive, often more structured and of more consistent quality, and not necessarily directly related to patients so can be accessed more easily.

On the clinical side, the adoption of AI will be more gradual and take longer, but Janak believes that we will start to see changes in about five years’ time.

The impact of the EMBA and “Electives for life”

Janak admits that he has always been very analytical and has a structured approach to doing things. He puts this down to his medical training and consulting background. The EMBA taught him to develop “more nuanced considerations of people’s motivations and thought processes”. This was an essential skill when writing the book. The EMBA also provided Janak with the theoretical knowledge he set out to acquire, which has proven invaluable in his day-to-day work when he is sometimes required to make decisions in the context of uncertainty.

Janak enjoys returning to Cambridge Judge for the EMBA Electives as it is a great way to reconnect with his fellow alumni and “keep up-to-date with the latest thinking across a range of topics. Some of the topics are not some that you would typically come across, so the electives serve the dual purpose of enabling me to catch up with friends and also get back into the learning environment.”

The post Artificial intelligence in healthcare appeared first on Cambridge Judge Business School.