More edits

Author: msjasteppe

articles/healthcare-apis/iot/concepts-machine-learning.md

@@ -29,33 +29,35 @@ The four line colors show the different parts of the data journey.
 
 :::image type="content" source="media/concepts-machine-learning/iot-connector-machine-learning.png" alt-text="Screenshot of the MedTech service and Machine Learning Service reference architecture." lightbox="media/concepts-machine-learning/iot-connector-machine-learning.png":::
 
-## Data ingest – Steps one through five
+## Data ingest: Steps 1 - 5
 
 1. Data from IoT device or via device gateway sent to Azure IoT Hub/Azure IoT Edge.
 2. Data from Azure IoT Edge sent to Azure IoT Hub.
 3. Copy of raw IoT device data sent to a secure storage environment for device administration.
-4. PHI IoT payload moves from Azure IoT Hub to the MedTech service. The MedTech service icon represents multiple Azure services.
+4. IoT payload moves from Azure IoT Hub to the MedTech service. The MedTech service icon represents multiple Azure services.
 5. Three parts to number five: 
    1. The MedTech service requests Patient resource from the FHIR service. 
    2. The FHIR service sends Patient resource back to the MedTech service. 
    3. IoT Patient Observation is record in the FHIR service.
 
-## Machine Learning and AI Data Route – Steps six through 11
+## Machine Learning and AI Data Route: Steps 6 - 11
 
 6. Normalized ungrouped data stream sent to an Azure Function (ML Input).
 7. Azure Function (ML Input) requests Patient resource to merge with IoT payload.
-8. IoT payload with PHI is sent to an event hub for distribution to Machine Learning compute and storage.
-9. PHI IoT payload is sent to Azure Data Lake Storage Gen 2 for scoring observation over longer time windows.
-10. PHI IoT payload is sent to Azure Databricks for windowing, data fitting, and data scoring.
+8. IoT payload is sent to an event hub for distribution to Machine Learning compute and storage.
+9. IoT payload is sent to Azure Data Lake Storage Gen 2 for scoring observation over longer time windows.
+10. IoT payload is sent to Azure Databricks for windowing, data fitting, and data scoring.
 11. The Azure Databricks requests more patient data from data lake as needed.
     1. Azure Databricks also sends a copy of the scored data to the data lake.
 
-## Notification and Care Coordination – Steps 12 - 18
+## Notification and Care Coordination: Steps 12 - 18
 
 **Hot path**
 
 12. Azure Databricks sends a payload to an Azure Function (ML Output).
-13. RiskAssessment and/or Flag resource submitted to FHIR service. a. For each observation window, a RiskAssessment resource is submitted to the FHIR service. b. For observation windows where the risk assessment is outside the acceptable range a Flag resource should also be submitted to the FHIR service.
+13. RiskAssessment and/or Flag resource submitted to FHIR service. 
+    1. For each observation window, a RiskAssessment resource is submitted to the FHIR service. 
+    2. For observation windows where the risk assessment is outside the acceptable range a Flag resource should also be submitted to the FHIR service.
 14. Scored data sent to data repository for routing to appropriate care team. Azure SQL Server is the data repository used in this design because of its native interaction with Power BI.
 15. Power BI Dashboard is updated with Risk Assessment output in under 15 minutes.