A lot has been said about how AI will revolutionize health care.
Indeed we have algorithms which can help radiologists assess X-Ray pictures or CT / MR results. We are also developing robots which can perform complex operations in the human body.
But there are also numerous other applications for AI techniques. Usually, they are not as sexy, but much easier to implement. And nonetheless, they still bring tangible benefits. Read on to get to know them!
Shorter waiting times for ambulatory care
In one of my previous posts, I described a story about how Mathematical Optimization, together with Lean approaches, contributed to a significant shortening of the patient waiting times in an Italian hospital. The results were indeed noteworthy – up to 94% reduction in patient waiting time!
This reduction was possible thanks to a smart combination of both techniques mentioned. Lean approaches enabled the identification and reduction of waste in the existing processes, and Mathematical Optimization helped to make the best use of existing resources every time a schedule was built.
Better kidney transplant choices
Stefanos A. Zenios and Xuanming Su from Stanford developed an algorithm which greatly improves the way in which kidneys are allocated to patients on a waiting list.
The effect: 10% more people could have access to a kidney for a transplant. Additionally, the share of discarded organs could drop from 11-15% down to 3%.
Can we afford not to have this implemented elsewhere in the world?
Less waste of blood components
Working with a blood inventory management system in Ontario, Li et al. developed a hybrid demand forecasting/ordering system for blood ordering.
Their strategy reduces the inventory level by 40% and decreases the ordering frequency by 60%. Moreover, shortages are virtually absent and there is only marginal wastage due to expiration – much less than with the existing practice.
More patient-facing time
For a community health centre in Vancouver, Zimmerman et al. came up with an algorithm to align the schedules of nurses with the patients’ demands.
It allowed for up to a 13% increase in time spent with patients and 10% more patients seen within one shift.
Improved doctors’ availability
For a Norwegian hospital, Klyve et al. came up with a metaheuristic approach to generate robust schedules for surgeons, accounting for emergencies and unplanned absences.
As a result, the problem of understaffing was brought to an end – the schedules were generated in such a way that a sufficient number of doctors were available on each shift. Moreover, the stability of the schedules improved – they did not change as frequently.
Stay tuned for future posts!
P.S. My post would be incomplete without a xkcd stripe. Here comes one.