A nurse has eight minutes before a patient goes home. In front of her: a printed discharge packet, a condition page in the portal, an animation the specialty clinic recommends, an inhaler trainer in a drawer, and a decision aid someone bookmarked last year. Format is not the constraint.
The real question is narrower, and it is the one worth designing around: which resource helps this person understand what is happening and what to do next, and how will anyone know whether that worked? HeyGen’s guide to patient education tools treats that as the organizing problem rather than a content-library problem.
The tool list is long because the jobs differ
The guide separates ten types: printed handouts and checklists, videos and multimedia, demonstrations and physical models, portals and EHR-delivered education, mobile apps and interactive modules, decision aids, action plans and trackers, trusted public libraries, group and peer education, and one-to-one teaching.
The useful comparison is job to format, not tool to tool. A handout that explains a procedure will not teach injection technique. A condition video will not help someone choose between two treatments. Decision aids are their own category because their job is weighing realistic options in a preference-sensitive choice, not following one recommended plan.
Print survives a dead phone battery and can be handed to a caregiver, but the guide is blunt that a packet handed over without discussion is documentation, not education. Portals standardize delivery and leave a reference behind, yet a link is a delivery channel, not proof anyone opened it or can act on it.
What the evidence for video actually supports
The guide cites a 2026 systematic review by Kathryn Jack and colleagues in the British Journal of Community Nursing covering 15 studies published between 2014 and 2021. Eleven reported improvements in patient knowledge or understanding after multimedia education. The review also found substantial variation across interventions and focused on short-term knowledge rather than long-term clinical outcomes.
That is a narrower claim than most vendor pages make. Multimedia can improve knowledge and understanding in many contexts; it does not replace checking whether a specific patient understood. The guide also keeps the business case separate from the outcome claim: hospitals face financial pressure through the Hospital Readmissions Reduction Program, but that pressure is not evidence that a particular handout, video, or platform reduces readmissions.
Five questions, in order
The selection sequence in the guide is goal, patient, access, quality, confirmation.
Start by naming which of four jobs the patient has: understanding information, learning a skill, making a decision, or self-managing a symptom. Then assess the starting point — what the patient already knows, what worries them, who is at home with them. Two minutes there often reveals the one thing the education should cover first.
Health literacy is handled as a default, not a separate thinner track: familiar words, key information first, logical chunks, active voice, useful headings. The guide’s internal rule is need-to-know before nice-to-know, with three prioritized points a patient can repeat beating twelve they skim. Language coverage is a separate decision from reading level, and the guide notes that machine translation alone does not meet the standard for medical content.
Access comes before commitment. Is the video captioned? Does it work with a screen reader? Can the patient open it after discharge, on their own device and data plan? A resource the patient cannot open is not education.
Confirmation is the step that gets skipped
The workflow does not end at “sent the resource.” Decide in advance whether you will use teach-back for knowledge or a show-me step for a skill, and who does it. That is what separates education from distribution.
With eight minutes, the answer cannot be teaching back everything. Verify the highest-risk point first — a medication change, a warning sign, device technique, or a next action that could cause harm if misunderstood. Move education earlier where possible and use the final encounter to confirm rather than introduce.
Before buying another platform, the guide suggests documenting what the health system already owns: available languages, reading levels, formats, portal and EHR delivery, automated assignments, comprehension feedback, reporting, review cadence, and post-discharge access. The incumbent landscape includes Epic-linked education, Elsevier PatientPass, Krames, Healthwise, UpToDate materials, and Wolters Kluwer’s UpToDate Educate. Capabilities vary by contract and implementation, so the missing piece may be interactivity, localization, analytics, or workflow configuration rather than another content library.
For teams building in this space, the same discipline applies to any AI-assisted pipeline that generates or translates patient material. The output is a draft until someone owns the comprehension check. If you are wiring generated content into a delivery workflow, the scoping questions in this post on granular authorization for agent access are a reasonable starting point for keeping that pipeline away from anything it should not touch.
A practical next step: pick one high-risk instruction your team sends today, write down who confirms it and how, and check whether the format you chose actually supports that check. The supplied guide does not specify how to measure long-term outcomes, so treat that as an open question for your own quality team rather than something the source settles.
Sources
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.