How Bot-Assisted Automation Can Help Providers and Patients

Identify The Benefits Of Automation

“If only we could clone you.” That compliment is often given to the most efficient and effective worker, and especially comes into play when a business has more work to do than their overall resources can accomplish. While cloning isn’t an option, some businesses are turning to robotic process automation.

Automation promises to streamline repetitive tasks by turning them over to computer algorithms, or “bots”. In healthcare, that may look something like the following.

Imagine a bot that reaches out to a patient before their upcoming clinic visit and does all the work of verifying contact and insurance information. Additionally, the bot may even ask the patient to describe the reason for their visit. Details about the present illness could be elicited and a list of potential diagnoses could be generated by the algorithm. Then, after the provider visits with the patient and reviews the suggested diagnoses and confirms the appropriate diagnosis, the bot could queue up appropriate treatment plans for the provider to easily select from.

The opportunities for automation in healthcare seem endless. After all, bots never get tired or sick like humans do. They will continue doing as they’ve been programmed to do as long as technical systems are up and running. But what might be the risks of jumping head-first into this new era of bot-assisted delivery of healthcare? What do humans provide that bots cannot?

Acknowledge The Risks Learning

One thing we humans have going for us is our ability to learn and grow. Each person brings their own evolving knowledge, unique perspective and new ideas to the workplace. As we interact, we learn from each other and improve our overall processes as a result.

While it is true computer algorithms can be built to learn from new data over time, this level of artificial intelligence is often not found in present day automation systems. Instead, most systems will generally continue running as originally built until a human being steps in and updates the underlying algorithm.

With the rapid evolution of medical knowledge and everchanging regulations, bots working in healthcare must be updated on a regular basis if they are to compete with the status quo.

Resilience

Another pro for humans is our resilience, especially when working as a group. A team made up of several people can make up for the intermittent deficiencies of any one person in the group. For example, if one person gets sick, the remaining members of the team can take on the absent member’s duties until they return.

On the other hand, if an algorithm that controls an army of bots fails because of a simple coding error or system downtime, the entire organization may suffer until the problem is identified and fixed. Centralizing and automating processes in some cases is akin to putting all your eggs into one basket. To prevent all the eggs from breaking, the underlying technology holding the basket must be extremely accurate and reliable.

Following software development best practices including robust testing and multi-person code reviews can greatly reduce the risk of algorithm errors. Additionally, building redundancies into the underlying technical systems can reduce the risk of system downtime. These are just a couple of the critical components of building a resilient digital environment.

Adaptability

A common complaint made by doctors is that of inappropriate alerting in the electronic health record (EHR). Alert designers seek an appropriate balance between over and under alerting. Then, where alerts fall short, healthcare providers are expected to adapt, using the information presented in the alert and their own clinical judgement in deciding how best to proceed.

Like humans, computer algorithms are also adaptable, at least insofar as they are built to handle a variety of possible scenarios. For some processes, automation algorithms may be built to accommodate all significant possible scenarios. In such cases, there may be no need for additional adaptation in real time. However, for complex processes, anticipating and building every possible scenario into the algorithm may not be feasible. In these cases, like with EHR alerts, human supervision steps are built in to provide the needed adaptability.

To best meet the needs of all patients, automated processes need to accommodate all significant patient and environmental variation. While much adaptability may be built into algorithms directly, steps built-in to leverage human adaptability, may also be required in many cases.

Thoughtfully Move Forward

Automation of healthcare processes promises to add a measure of increased efficiency and quality. To meet that expectation, associated risks need to be identified and mitigated. We need not shy away from moving in the direction of robotic process automation but let us proceed wisely, taking the necessary steps to combine the power of computer algorithms and human review to ensure optimal outcomes are achieved for patients.