As a chief information officer and Harvard professor for two decades, I’ve worked in both the digital health innovation ecosystem of Boston and the policymaking world of Washington, DC. During this time, we’ve computerized records, enhanced security and built population health tools.

In 2018, we are now on the cusp of something much greater. Although I cannot predict the next five years, I can predict the next six quarters. Here are five projections of what’s next in healthcare information technology.

1) Clouds Are On The Horizon

A few legacy applications will still require local hosting, and some healthcare devices with high bandwidth/low latency requirements might require a hybrid of local hosting and cloud; however, most healthcare applications will migrate to the cloud. Some will be subscription services that require no capital investment and no software licensing. Agile CIOs will procure these services rather than provision them, as meeting the needs of customers will shorten implementation times and lower total ownership costs. The days of building data centers for hospitals and healthcare organizations has passed.

Although it may take a bit longer, I predict the traditional Windows desktop will also disappear, replaced with Chromebooks running cloud-hosted productivity tools.

2) The Status Quo Will Get a Mobile Makeover

Eighty percent of all website traffic at Beth Israel Deaconess Medical Center (BIDMC) is now viewed on mobile devices. Both patients and providers want to access records, schedule services and manage medications on their phones. Traditional “portals” will be replaced with apps that integrate internet of things (IoT) data, where patients self-report their symptoms and outcomes and electronic health records (EHR) data assists patients and families with care navigation. Telecare/telemedicine on mobile devices will be used to keep patients healthy in their homes instead of requiring in-person visits, time-consuming commutes and expensive parking. Just as Amazon has changed our shopping experience, innovative healthcare systems that embrace new mobile and IoT workflows will displace the status quo of health delivery.

3) Machine Learning Will Leverage Past Data for Future Success

Although some technologies are speculative, machine learning in healthcare works today. I predict many traditional analytic methods, such as business intelligence or big data, will be replaced by machine learning methods that leverage the experience of millions of past patients to predict successful treatment plans for new patients. Examples we’ve experimented with include:

• Optimizing operating room scheduling
• Forecasting length of stay by patient (and then working backward to get them home on time)
• Determining who is likely to miss an appointment

A patient being able to say, “Alexa, schedule a routine appointment with my cardiologist,” would be a game changer.

All of these technologies require curated, high-quality data from the past. High-quality data sets for training machine learning algorithms will become increasingly valuable, since much of the data available today in healthcare is incomplete, inaccurate and unstructured. Artificial intelligence will not replace clinicians; it will augment them. Although algorithms can identify patterns, they cannot apply emotional or human factors. If your doctor can be replaced by a computer, then he or she should be.

4) Ambient Listening Goes (Even More) Mainstream

“Alexa, ask BIDMC what my care plan is for today.”

This is the next logical step after mobile apps. As sight and dexterity diminish in aging populations, the use of voice becomes increasingly more important. By the end of 2018, I predict Amazon Alexa will be covered by business associate agreements so that healthcare organizations can design Alexa skills that interact with EHRs and practice management systems while managing privacy.

A patient being able to say, “Alexa, schedule a routine appointment with my cardiologist. What’s the next available date?” would be a game changer for a healthcare system. We’ll even be able to use sentiment analysis to assess voices for stress, anxiety and depression.

I also anticipate that providers will increasingly want to use ambient listening in their daily workflows. Today’s EHRs impose significant burden on clinicians. If documentation, ordering and patient flow can be controlled via ambient listening, efficiency improves and clinicians can practice at the top of their license.

5) Blockchain Will Not Replace EHRs

These emerging technologies depend upon patient consent for data sharing. Patients must trust the applications and infrastructure we build if they are to embrace it. Blockchain can help in two ways. It can host patient consent preferences in smart contracts that any application can access, clarifying how data can flow among stakeholders while respecting patient privacy preferences. Blockchain can also provide audit trails that ensure integrity of a medical record and track where data was exchanged. Blockchain will not replace EHRs—it’s not a database—but it will help build trust for health data sharing.

These five predictions may soon become five industry realities, as digital health innovation is well-funded and many new startups are well on their way to delivering novel products and services that layer on top of today’s existing EHRs, without the need for expensive EHR or infrastructure replacement. Let’s see what unfolds next.

This article first appeared on the Oliver Wyman Health website.