Configure built-in scenarios
To configure the built-in scenarios of an Azure Health Bot instance, expand the Configuration menu.
The Configuration menu contains detailed configuration items that can control every aspect of the Health Bot instance functionality and apply to every scenario. The configuration items are grouped in Medical, Compliance, and Conversation categories.
Medical protocols
On the Medical protocols tab, you can configure the behavior of the built-in triage and symptom checking and medical information scenarios.
Let's start with the configuration of the built-in medical Triage and symptom checking scenario.
Content providers
Depending on your needs, you can select Infermedica or Capita as your content provider. There are different settings options for each one. For the purposes of this module, only the Infermedica settings are detailed.
- Infermedica provides an advanced triage engine based on clinically validated probabilistic models.
- Capita uses a sophisticated rule-based decision tree, eliminating urgent cases in as few questions as possible.
Under Infermedica Settings, you can configure the following parameters.
General
Under General, you can edit the:
- Message when returning to triage from an interrupting scenario.
- Third-person disclaimer text.
Pregnancy
Under Pregnancy, you can:
- Turn on or turn off prompts for pregnancy status.
- Specify the age range to ask about pregnancy.
Gender
You can specify the question to ask the user about their gender.
Preassessment
You can configure questions for symptom collection.
Assessment
You can configure symptom checking and triage questions.
Triage results
You can configure title text for suggested care and possible causes.
On the Medical Information tab, you can edit the responses for questions about conditions, symptoms, and online resources.
Compliance
Under Compliance configuration, the following topics can be changed.
Privacy
Configure how users can manage their data.
Security
Configure the security settings for your bot instance. For example, you can change the timeout duration and message and require user authentication.
Terms and consent
Configure terms and consent to meet legal requirements. For example, if you've enabled the consent mechanism, users are prompted to agree to your Terms of Use and Privacy policy before they can interact with the bot.
Conversation
Under Conversation configuration, the following topics can be changed.
Interactions
Configure built-in scenarios that interact with your users. For example, you can configure:
Default error or retry messages.
A default message when returning from interrupting scenarios.
A default reply for utterances that aren't understood.
You can also change the automatic welcome message that appears when users first enter the chat. Or, you can use a custom scenario as an automatic welcome scenario. A custom welcome scenario overrides the automatic welcome message, but note that a custom welcome scenario doesn't work with multi-step scenarios.
Navigation
Configure the built-in scenarios that help users navigate conversations. Examples include the cancel, start over, and help commands.
In the Help section, you can control every aspect of the help menu. You can add or remove items and edit the description of each menu item.
Spelling
Health Bot provides a Spelling resource, so that you can configure the built-in spelling correction available in your bot. The service passes all utterances through a spell check before being processed by the natural language understanding model. If the Spelling service is enabled, user utterances are spell checked if they aren't understood in their original spelling. For example, a user might have spelling mistakes.
On the Spelling tab, you can configure the:
Spelling verification message that displays when the speller has corrected an utterance.
Message that displays when the speller has corrected an utterance but user intent wasn't understood.
Message that displays when the user rejects spelling verification.
Human handoff
Health Bot can be used to provide users with chat support from live agents. In some cases, it's preferable to transfer users from a bot interaction to a human conversation. For example:
- A triage flow can be handed off to a doctor or nurse.
- Support agents can resolve issues that require nongeneric assistance.
- The user can ask to be transferred to a human.
On the Human Handoff tab, you can enable handoff to human agents' functionality and configure related parameters. For example, the user and agent timeout and messages that are displayed when a user is waiting for an agent to connect or when an agent connects to a conversation. Handoff to live agents by using Microsoft Teams and integration with Dynamics require more configuration.
