AI: 4 practical paths to better care at lower cost
September 28, 2018
Healthcare systems, across the developed and developing world, face enormous challenges. The growing elderly population, shifting burden of diseases towards chronic conditions, shortage of healthcare workers and need to protect lower-income groups require a smart response. “Business as usual” is not an option. Healthcare delivery must be efficient, lessening the burden on governments and consumers.
My view is that changes and improvements in technology, especially AI, have made alternative strategies and healthcare delivery models possible. What is needed is a clear vision to experiment with new solutions to deliver effective care at an affordable price. Here are my four practical ways to reduce costs while improving care:
Reduce non-value-adding activities
One of the big challenges for hospitals is the amount of administrative burden placed on nurses and doctors that distracts them from their core function, which is taking care of patients. As evidenced by Accenture’s 2018 Consumer Survey on Patient Engagement, enthusiasm for AI to take over administrative functions is high. AI can take over many of these non-patient care activities. Accenture’s analysis shows that administrative workflow assistance—such as using voice-to-text transcriptions for writing chart notes, printing prescriptions and ordering tests—can result in 51 percent work time savings for nurses and nearly 17 percent for doctors.i
Predict the population at risk
Around five percent of the patients account for 60 percent of healthcare costs in the developed world.ii Patients with chronic diseases and co-morbidities place the greatest demand on healthcare systems. Anticipating which patients are moving towards this status, and designing interventions to prevent unplanned admission risks, can considerably enhance savings and reduce the burden on hospitals.
AI can play a significant role here. For example, Deep Patient—the in-house AI system at New York’s Icahn School of Medicine at Mt. Sinai—can predict risk factors for 78 diseases based on an analysis of electronic health records.iii As demonstrated by Accenture’s pilot with the La Fe hospital in Valencia, Spain, being able to identify patients with a heightened risk of avoidable hospitalisations can result in drastic reduction in hospitalisation cases and significant cost savings. A year-long clinical trial with 500 patients showed a 22 percent improvement in quality of life, and a drastic reduction of up to 79 percent of hospitalisations and emergency room visits. Extending this program to 800 patients freed 21 bed days during the clinical trial. Estimates suggest that extending the program to 4000 patients could deliver net savings of around seven percent of the total health expenditure to the La Fe Health Department.
Augment the workforce
AI has the power to bridge the healthcare worker demand-supply gap of. Molly, developed by the medical start-up Sense.ly, is a virtual nurse tracking patients’ everyday care. While current virtual assistants look at single assistive application, future assistive agents will be able to combine patients’ complete genome sequencing with a full body scan and detailed medical check-up to offer complete preventative care. The entire process would help identify cancer or other chronic diseases in their very early stages and identify risk metrics based on a user’s genetics and environment.
Bringing AI to the operating room to perform minimally invasive surgeries can not only enhance precision but also lower the cost of treatment, with faster recovery and reduced length of stay in hospital. It can potentially convert inpatient surgeries into outpatient surgeries with lower risk of hospital-acquired infections. Mazor Robotics, for example, is using AI to assist minimally invasive surgical operations. Mazor’s spinal surgery robot arm can guide the surgeon’s instruments with extreme precision based on a 3-D computerised planning system that indicates where a surgeon should place implants.iv
Shift the locus of care to home
New AI solutions can continuously monitor a patient’s vital signs in real time using wearable devices, various digital technologies and powerful analytics. Such solutions require significantly lower capital expenditure than traditional bedside monitoring equipment, making them a cost-effective way to keep a constant eye on patients, even as they move out of intensive and critical care units. The system continuously analyses vital signs from the digital plaster/device, and alerts appropriate physicians and staff via their digital devices when a patient’s condition is likely to deteriorate. Without adding costly equipment, busy healthcare professionals’ time is optimised and limited beds in units with higher levels of care are used more effectively, all while helping to improve the outcomes of care. This platform is now being integrated to support two Indian partners with robust and low-cost solutions.
As an approximation, a decrease in the average length of stay of patients by four hours in a 275-bed capacity hospital can translate into increasing the physical capacity by 10 beds.v However, early discharge of elderly patients from the hospital can stumble on multiple factors, such as lack of caregivers at home or lack of confidence in being able to manage themselves. Mapping the decision-making process of elderly patients and being able to predict factors that increase their length of stay could help hospitals take the necessary steps to alleviate the problem. Even more significantly, using AI to help the elderly live independent lives could be a big step in reducing the burden on both hospital systems and caregivers. The Japanese government is leading initiatives to design robots that can help elderly people live independent lives by providing mobility, toilet, lifting and bathing assistance. Nursing care robots can help reduce the burden on workers and prevent them from quitting.vi
The potential applications of AI extend well beyond robot assisted surgeries, virtual nursing assistants and administrative workflow assistants to include fraud detection, dosage error reduction and cybersecurity. A comprehensive adoption of AI will, however, require seamless integration of data sources as well as ecosystem play to ensure ethical use of AI.
But I think the writing is on the wall—applications of AI in health are happening here and now. Embracing these applications make cost-effective, high quality care more possible. AI in its various forms—such as predictive analytics, connected devices and machine learning—can offer effective solutions to provide care anywhere anytime at affordable cost. Do you agree? Let me know.
i Matt Collier, Richard Fu, Lucy Yin, “Artificial Intelligence: Healthcare’s New Nervous System,” 2017,
ii Accenture, “Predictive Health Analytics,” 2017, https://www.accenture.com/t00010101T000000Z__w__/gb-en/_acnmedia/PDF-53/Accenture-Health-La-Fe-Credential-Predictive-Health-Analytics-Model.pdf#zoom=50
iii Accenture Technology Vision 2018, “Redefine Your Company Based On the Company You Keep,”
iv Matt Collier, Richard Fu, Lucy Yin, “Artificial Intelligence: Healthcare’s New Nervous System,” 2017,
v Hewlett-Packard Company, “Healthy, Wealthy and Wise,” Technical White Paper, December 2012.
vi Japan to Create More User-Friendly Elderly Care Robots, Robotics Trends, 20 Nov 2015,