AI Scribes in Public Health: Reimagining Workflows Around Care

Table of Contents

Introduction #

If you’ve worked in public health, you already know this feeling.

A single client interaction rarely stays in one place. Notes are written, then rewritten. Information is entered into one EMR (Electronic Medical Record) system, then entered again into another (e.g., iPHIS). Investigation forms, whether developed by Public Health Ontario (PHO) or by local public health units for specific diseases, must also be completed. What begins as a conversation can become a series of fragmented entries across platforms, pulling staff time and attention away from listening, care planning, and supporting people and communities.

AI scribes can help shift that attention back.

An AI scribe listens to conversations (in-person or virtual), extracts key information, and generates structured clinical or case notes automatically. The promise is not just about automating notetaking, it’s about improving care. The opportunity is to use AI scribes in ways that help staff stay more present with clients, improve service quality, and reduce the administrative pull away from the interaction.

Why does this matter? Public health encompasses surveillance, prevention, outreach, and coordination beyond clinical care. Better documentation can improve data quality, reporting, decision-making, and population health outcomes. The most meaningful impact starts earlier: in the quality of the client interaction itself. At Wellington-Dufferin-Guelph Public Health (WDGPH), we’ve been exploring this challenge firsthand. Over the past year, we’ve piloted AI scribe technology across multiple public health teams. The goal was to understand how it behaves in real, day-to-day public health work and where it may allow teams to work differently.

Reimagining Public Health Workflows with AI Scribes #

AI scribes are not “one-size-fits-all”. Their value depends on whether organizations are willing to look beyond simple automation and ask how the work could feel better for both staff and clients. Transformational integration requires outside-the-box thinking: stepping back, questioning old handoffs, and testing new ways of capturing information without weakening the human connection.

Below are examples of workflows identified in WDGPH for piloting AI scribes. Think about your own teams: which interactions involve multiple systems, repeated note-taking, or long conversations that later need structured documentation? Those are the places where AI assistance may create space for better care and a smoother staff experience.

Example workflows:

Infectious Diseases investigations (e.g., Salmonella disease investigation)
IDA home visits for children’s health services
Immunization-related inquiries over a call
Sexually transmitted disease investigations
TB or chronic disease management
Dental hygienist appointments

Choosing the Right AI Scribe Product #

There are many AI scribe products on the market, each with its own strengths. Some focus on clinical note generation, others specialize in ambient listening, and newer tools can handle form filling and downloadable attachments. When choosing a solution, it’s important to ask:

Does it support multiple custom templates?
Can it support different ways of working across teams?
Does it integrate with the systems staff still need to use?
Does it help staff stay focused on the client instead of the screen?

The “best” AI scribe is the one that helps your team rethink the work, reduce unnecessary duplication, and improve the service experience for staff and clients.

For Ontario health organizations, Supply Ontario has established a new Vendor of Record (VOR) arrangement for Artificial Intelligence Solutions – AI Scribe. This arrangement includes multiple vendors who have been pre-qualified through a competitive Request for Bids (RFB) process, making it easier for organizations to find a solution that meets their needs. You can explore the VOR here: Supply Ontario AI Scribe VOR.

Privacy and security considerations #

Public health data is sensitive, so any AI scribe solution must meet strict privacy and security requirements. Key considerations include:

Regulatory compliance – The tool should comply with local privacy laws, such as Personal Health Information Protection Act (PHIPA) in Ontario, and ideally have System and Organization Controls 2 (SOC 2) certification for security assurance.
Data protection – Notes and information must be encrypted both in transit and at rest, with secure deletion protocols.
Clear data policies – Vendors should have transparent data retention and de-identification policies.
No unauthorized data use – Data should never be used for model training or secondary purposes without explicit consent.
Role-based access – Access should be controlled and limited according to staff roles to maintain confidentiality. Ensuring these safeguards are in place protects both your clients and your organization, while allowing AI scribes to be safely included in public health practice.

Example of AI Scribe Template #

Below is one version of a custom template designed for In-Depth Assessment (IDA) home visits within children’s health services at WDGPH. It can be customized into a goal-based structure for more comprehensive data capture. That flexibility creates a governance challenge: AI scribe tools do not currently provide built-in version control at an organization-wide level, leaving no centralized way to track, manage, or roll back changes across teams.

Figure 1. AI Scribe Template for IDA Home Visits

Date and time of interaction: (YYYY-MM-DD at HH:MM)

AI Scribe consent obtained.

D - Data:
(Summarize the client’s current concerns, challenges, or issues shared during the visit. Focus on what was subjectively reported by the caregiver and what was objectively observed by the PHN. This section should reflect the specific problems or needs identified during the interaction and will be addressed in the Action section. Use a maximum of 6–7 bullet points, each starting with a hyphen (-), and write a concise 3–4 sentence narrative per point. Prioritize clarity, keep it concise, and maintain a narrative that tells the story. Only include information if it has been explicitly mentioned in the transcription, otherwise print ‘N/A’.)

A - Action:
(Using hyphen (-) as bullet points, document how the PHN addressed the concerns listed in the Data section. Include the actual interventions and education provided — not just the technique used (e.g., detail what was taught, modeled, or discussed). Note the use of therapeutic strategies (e.g., motivational interviewing, role modeling), stages of change, and how the intervention aligns with relevant FSP goals. Group actions logically rather than repeating for each concern. Keep entries concise but informative. Only include information if it has been explicitly mentioned in the transcription, otherwise print ‘N/A’.)

R - Response:
(Using hyphen (-) as bullet points, record the caregiver’s reaction to each of the actions/interventions. Document emotional responses, verbal feedback, level of engagement, expressed motivation, or requests for follow-up. Keep responses concise. Only include information if it has been explicitly mentioned in the transcription, otherwise print ‘N/A’.)

What did we observe in testing? #

One of the most significant challenges in testing was template management. Without built-in version control, individual users are often responsible for creating and maintaining their own templates, which can lead to inconsistencies, duplication, and variable documentation quality. The challenge increases across disease areas, where requirements are highly specific. For instance, a Salmonella investigation template cannot be directly reused for other enteric diseases due to differences in reporting requirements, data points, and clinical context. This underscores the need for stronger governance over template creation, updates, validation, and retirement.

There are also important considerations around accuracy and adoption. Like all generative AI tools, AI scribes can produce incomplete or incorrect information. Staff must review and validate notes before using them, in line with professional and regulatory expectations. Adoption also takes time. Moving away from traditional documentation practices requires adjustment, clearer standards, and confidence that builds through regular use. Staff with lower digital literacy may need more support, which can influence consistency across teams.

Technical limitations can add further barriers, including inconsistent experiences between browser-based and web applications, audio compatibility issues, and reliance on specific hardware. Limitations in role-based permissions and account management can make onboarding, offboarding, and staff reassignment more difficult, while differences in digital familiarity can affect consistency of use.

Strong change management is therefore essential. A dedicated lead or team champion, ideally with subject matter expertise, can support template development, share best practices, and build confidence. These challenges also raise broader considerations around equity and accessibility. All staff need enough support and training to benefit from the technology in a sustainable way.

Measuring ROI: Beyond Time Savings #

Time savings are often the first benefit associated with AI scribes, particularly in long unoptimized workflows. The deeper value is enhanced service: better focus during the interaction, stronger connection with the client, and less cognitive load for staff after the visit or call.

Key Metrics to Consider #

To understand the true impact, it’s important to look at a broader set of measures:

Reduction in documentation time – Are staff spending less time when using an AI scribe compared with a manual documentation process?
Staff satisfaction and burnout – Is the tool helping reduce fatigue and improve day-to-day experience?
Client focus and connection – Are staff able to stay more present during the interaction?
Ease of use and adoption – Are teams using it consistently, and does it support better ways of working?
Improvement in data quality – Are notes more complete, consistent, and reliable?

Together, these indicators provide a more complete picture of whether the AI scribe is improving care, staff experience, and documentation quality.

Workflow Suitability Matters #

Workflows that involve high-volume, repetitive documentation tend to see the clearest return, as AI scribes can reduce duplication and manual effort. Complex or highly variable cases may see different benefits, especially when the value comes from helping staff listen more fully, organize information more clearly, or follow up more consistently. Taking the time to match AI scribes to the right service moments is key to maximizing impact.

Conclusion #

Success with AI scribes depends on more than introducing new technology. Teams need to step back, reimagine how information moves through the work, establish clear governance and standards, and address privacy, security, and equity considerations early. Ultimately, the value of AI scribes in public health lies in helping staff spend less energy on documentation and more attention on the people and communities they serve.