Telehealth: Lessons from the Rideshare Model

As telehealth popularity surges, clear evidence shows that this new delivery model has great upside potential when things go right. It also is clear that there are tricky issues to address and significant downside consequences if things go wrong with attempts to deliver on-demand telehealth services at scale. Uncertainty and variability in patient service needs and provider availability cause challenges in matching demand to supply. A key to success in capturing the upside potential and avoiding downside risks is the use of predictive analytics, which .

Medecipher already has been applying successfully in the Emergency Department (ED) setting. Medecipher’s ED experience and solution is positioned to provide significant value for the telehealth setting. 

We can look to a tech-intensive older cousin that serves a very different industry sector — ridesharing — to see how the important ability to accurately predict volumes in an irregular flow model contributes high value.

First, let’s take a quick look at the dynamics of ridesharing.

For Uber, success depends upon effectively interweaving a pair of exceptional capabilities. Not only do ridesharing firms need to connect supply and demand, but they also need to effectively deal with the complexities of managing both aspects of their two-sided supply-demand model. 

  1. Uber must provide its rider customers with a sufficient supply of drivers, and 
  2. Uber also needs to provide its drivers with a sufficient supply of riders.

The effective balancing of these, in real-time, is crucial.

  1. If riders’ wait time is too long, due to an insufficient driver fleet, they will abandon transactions and seek alternative vendors or methods of transportation.
  2. If drivers don’t get consistent enough rider volume, the high level of downtime will discourage driver loyalty, and it won’t be worth their time to be available for rides. 

Entrepreneur and investor David Sacks depicted what he called “Uber’s virtuous cycle” in the below illustration:

The model applies fairly directly to the telehealth instance, though the dynamics and expectations on both sides of the equations shift in some important ways. As we replace riders with patients, there is (at least so far) a more relaxed expectation on the “pickup” time. But provider expectations are high — for credentialed care providers, the value of their time is high, and the availability is constrained. 

Financial and operational success for telehealth providers relies on effective queuing and demand-supply matching. Specifically, two sides of the equation match provider availability to patient demand

  1. Provider loyalty for consistent availability relies on right-sizing demand to be “on-call” to retain providers — need to predict demand volume for preferred providers to be on-call (e.g., providers would be willing to remain on call if they see 1 patient/hour)
  2. Patient satisfaction to drive utilization and reduce leakage relies on right-sizing supply for low wait times to be seen (e.g., patients exit the queue/go elsewhere if waits > 2 hours)

With these considerations, leveraging a reactive demand-supply model won’t work. It needs to be predictive, to allow providers to plan their availability at times that patient demand will justify it. Current prediction precision is lacking, and it’s an expensive problem. Telehealth providers utilize secondary incentives (such as provider loyalty bonuses) to compensate for demand inconsistencies and resulting dissatisfaction. By optimizing staffing levels, telehealth providers can save money on such incentives, and increase patient throughput, satisfaction, and service volume (patient census).

This is where Medecipher can make a meaningful impact. The up-and-coming health tech company was recently awarded supplementary grant support from the National Science Foundation (NSF) to adapt product forecasting models for estimating the telehealth staffing needs. If you’re interested in partnering with us to explore this application, contact us at