More detailed answer:
I'm going to start by not even touching AI and instead focusing on payor incentives and fee schedules:
- Medicaid
- Fee schedules are set, but vary by Medicaid provider type. MCOs have some flexibility to go up or down or agree to alternate payment models, but for the most part provider payment is based on the Medicaid fee schedule for whatever provider type they're under.
- Medicare
- Fee schedules are set and vary by geography
- Commercial (Employer + Marketplace; looping in Medicare Advantage too)
- They may have a default fee schedule, but it's negotiable. Larger networks and/or ability to handle needs of special population groups on the provider side can help negotiate better rates.
- Self-pay
One of the challenges with AI savings getting passed on to patients in the short term is that the payers/government still have to make the fee schedules reasonable to compete for non-tech-enabled providers. They and state licensing bodies can try (and sometimes succeed) in differentiating the VC-backed, tech-enabled, for-profit groups from the rest by requiring in person services to access certain fee schedules, disallowing for profit entities from engaging in certain care, or jacking up compliance requirements, giving the more traditional clinics access to better rates, but there are some legal, operational, and practical challenges in doing that.
If a tech-enable provider (or even a traditional provider) finds some AI use that saves them 10% of operational costs, it's not like they're going to start charging below the fee schedule they have access to. If anything, it may free up cash to let them hire more, spend more to acquire patients, enter new markets, etc. AI efficiencies are likely to help the tech-enabled groups scale and increase their negotiating leverage to increase commercial fee schdules.
There's also going to be a "cost of doing business" component to successful AI uses. As in, the person providing the AI is going to want to get as much as possible eventually. They may underprice for a while to get marketshare, but they will press up the price at some point, especially if providers are heavily relying on the service, have invested in integration, and don't have good alternatives.
For your specific use cases:
- Billing
- Will keep this separate from RCM and focus on coding and patient responsibility. AI already increases upcoding, often through scribe software, and that's a key value prop for those companies when approaching practices.
- I expect AI to play a role in trying to improve collection rates and speed on patient responsibility
- Don't expect either of these uses to decrease costs.
- Scheduling + Call centers
- I'm grouping these because AI is taking on some similar tasks and replacing similar skill leveled employees. There will be some cost savings here, but it's a more complex problem than most realize. I expect the cost savings to be realized by large, tech-enabled providers and for non-tech-enabled providers to pay third parties with generally poor results and ROI. I don't expect this to translate to patient savings.
- Insurance processing + RCM
- Combining these because I see AI playing a role in an escalating battle. As insurance company pressures continue to tighten, they'll have less and less incentive to invest in improving their processing systems. They will be incentivized to put up more barriers to processing (underfunding contract loading staff, slow cred timelines, old systems/software, increasing provider audits/documentation review). Payors may also use it to identify providers to drop. They'll balance all this with litigation risk.
- RCM may use AI to assist with day to day work, finding patterns, etc., but I see the AI value here to come from using software that's making payor pattern observations across tons of providers and having AI use those data to build payor-specific billing rules and adapt them over time as payor patterns change.
- There'll be some efficiency from providers who can take advantage of the AI in RCM, but they'll be combatted by payors trying to pay slower and less. The ones who can't take advantage of these patterns will be at risk. I see this as an evolution of the battle and not something that'll drop costs. It'll likely make it harder on providers, and providers are already struggling with this stuff.
- Documentation
- Assuming you're asking about EHR documentation. There's already and will continue to be some benefit to provider experience with this. It'll vary based on note structures and practice types. I don't expect this to lower costs for patients.
- There'll be some patient value from AI-assisted record access, letters, etc., though I don't expect it to lower costs
I do expect some AI-driven efficiencies that help the self-pay group, though I expect this to be dampened a bit by ongoing regulation pushed by the states (who are influenced by incumbents). Even if there are good self-pay options, people paying for them are still likely to have insurance and be paying for that regardless.
Long term, there's hope for cost containment, but it won't be easy. I do expect some improvement in user experience across most parties in healthcare. Just don't see costs going down in the short term. Some healthcare operational costs will drop in certain areas, but I don't see those getting passed on to patients.