“What derails AI (artificial intelligence) deployments?” This was the exact question guest speakers: Sami Manjure, Co-Founder and Chief Executive Officer of KenSci and Clifford Goldsmiths, Director Business Development & Strategy, Microsoft, tried to answer at the recent AIMed webinar took place on 25 June. 

Ways to make the most out of AI 

Manjure said AI in healthcare may sound sexy and polished, but it is a mean to an end. Often, if the reason for adopting AI is unclear, there is no chance for it to succeed within the system. In general, Manjure believes AI has the capabilities of improving population health, enhancing the overall care experiences, reducing per capita cost of medicine, and expanding the productivity of clinical teams. 

In the past, the results of AI were moderate because there was no appropriate tool to carry it out. Thus, in order to make the most out of AI, there is a need to improve infrastructure and exercise flexibility. Goldsmiths said, with Cloud technology, health data and public data coming from wearables and other devices could now be safely stored and solutions can be built directly on it. However, tapping onto the Cloud only digitized data, there remains a need to ensure a developed system that can engage everyone at work. 

Goldsmiths added the old patient paradigm is reactive and cyclical. Patients are accepted to treat their disease and it depends on the patients to come back after they are gone. In the new patient paradigm that is to be facilitated by AI, it will be preventive, connected and continuous. There are wellness plans and remote monitoring in place, to keep patient connected to the healthcare system.  

The three successful used cases 

The guest speakers cited three examples from the US, UK, and Singapore, to illustrate use cases that were brought to success because of AI. In the US, AI was employed for early detection; to predict the onset of disease, mortality, and readmission. As Manjure cited, a 76-year-old African American male, who was trapped in a cycle of admission and readmission till he dies. If there is a machine learning (ML) model in place to predict in advance, the rate of readmission and mortality, resources could be better utilized to keep the gentleman out of the care system while prolong his life and better the quality of care. 

In the UK, the opportunity was to reduce the annual patient’s emergency hospital admission, especially among high-risk Chronic Obstructive Pulmonary Disease (COPD) patients. With the use of AI and ML to help recognize patterns in patients’ conditions and provide early warnings to aid in intervention and prevention, the National Health Service (NHS) Scotland can now remotely monitor patients’ symptoms, physiology, and treatment at home. The ML algorithms also offer caregivers clinical decision support to assist high-risk patients and to deliver a continuous digital healthcare service. 

Singapore declared a war on diabetes and the country is using AI at a national level to combat against the chronic disease. As many as 1.5 million people who are predisposed with diabetes were given wearable devices. These data were embedded into ML model to motivate individuals to change their lifestyle and increase exercise routines, in order to build a healthier population. 

 To instill confidence 

Goldsmith noted most of the time, it is not just about having an AI model but also getting the right people at the right time in the right way. There is a need to have a foundation in place for AI to work. Therefore, it is important to instill confidence in those who are about to come on board or those who are already on board. Manjure commented transforming data into healthcare with impact is a complex team sport. 

The Chief Medical Officer could act as the gatekeeper in ensuring the AI model is working towards solving the intended problem, The Chief Financial Officer or the Chief Operational Officer could be the critic, to improve the general deployment of the model and the company, and last but not least, the Chief Intelligence Officer shall be the navigator to decide on where to move on next. 

As Goldsmith cited the projection put forward by Frost & Sullivan, the healthcare AI market will increase at a compound annual growth rate (C AGR) between 47% and 50%, reaching $36.1 million by 2025 conservatively. Therefore, the potential is there, it just needs the right kind of mindset and management to turn it into a reality. 

The webinar is now available for revisit. Do follow us on Twitter, Instagram, Facebook and Youtube for more event updates.

Author Bio
synthetic gene empathy chinese artificial intelligence data medicine healthcare ai

Hazel Tang

A science writer with data background and an interest in the current affair, culture, and arts; a no-med from an (almost) all-med family. Follow on Twitter.