Soothing Behavioural Techniques
System Re-Learning
The accuracy of the system will depend on the amount of user self-reporting and correcting of irrelevant adaptive responses.
PROJECT SUMMARY
The Problem: Currently available apps for menstruation tracking are leaving users frustrated and anxious. The lack of focus paid to the psychological symptoms of menstruation in current Period Tracking Apps is a gap in the market.
Project Type
End-to-end app & Advanced HCI
Industry
Health & Wellbeing
Tools
Figma, FigJam, Microsoft Suite
The Solution: An emotionally responsive and informative mobile app utilising physiological and ovulation data and an affective computing system to detect patterns of mental health.
Discovering the problems
Secondary research revealed the following core issues with the most popular apps available to users:
-
Lack of explanation or reassurance.
-
Negatively impacting user mental health
-
In-accurate cycle predictions relying on user input
Understanding the emotional framework
In this project, I explored how affective computing can be applied to support individuals experiencing stress and anxiety associated with PMS and PMDD. My focus was not on eliminating these emotions, which are hormonally driven, but on investigating ways technology might lessen their intensity and reduce their hold over the user.
A key part of my method was designing user flows that required active self-recognition of emotional states. This decision was shaped by my aim to not only enable AI-driven emotional support but also to encourage the development of emotional intelligence through self-reflection and regulation.
Research
Multi-Modal Fusion
The project applied affective computing to detect, monitor, and respond to users’ emotional states, offering real-time, personalised support for anxiety regulation.
The system combined physiological data, historic cycle insights, and user self-reported input through a multimodal hybrid fusion strategy to improve accuracy and adaptability.
Smart Ring Technology
The project also planned to integrate the OURA Ring and it’s Cycle Insights algorithm, allowing the system to account for irregular cycles with high reliability, while fallback to calendar-based tracking ensures continuity if physiological data is unavailable.
System transparency was also prioritised within the system so to notify users of any reduction in accuracy or changes in readings.
Defining the User
Personas and scenarios
Personas and scenario storyboards were created to allow for better contextualisation and empathy of the users' needs, wishes and constraints in relation to the problem.
Defining the Solution
From Problems to Solutions
Without accurate insights into their body and it's symptoms, users jump to conclusions and make assumptions based on misinformation. This results in an increase of negative emotional responses, which are already common in users with PMS & PMDD. Without support or guidance, hormonal anxiety can escalate and further impact the user. From this the individual will fall into a pattern of emotional avoidance and self-destruction.
Creating the Prototype
App Structure
To ensure all steps within the adaptive system were included, a full information architecture of the app was developed.
The IA allowed for user flows to be structure and prioritised based on affective experience. Within the architecture, key affective computing user flows were indicated.
Prototype
Onboarding
Soothing Behavioural Techniques
Correctly recognising anxiety
Evaluation
Evaluation of Affective System
A study plan was created to evaluate the accuracy of the systems recognition of anxiety in it’s users using a Confusion Matrix and collected during the longitudinal study.
Adaptation would be measured through user interaction with the system over the longitudinal study and sentiment analysis of data gathered in User Diary logs.
For it’s user-centred metric, this study will gather emotional responses (self-reported) within the app during the longitudinal study. This will then be corroborated by post-test questionnaire and focus groups.
What I've Learned
.png)
Designing for Emotional Context
I learned how to incorporate affective computing principles to recognise and respond to user emotions within a sensitive domain.
I gained appreciation for balancing empathy with practicality—how much emotional adaptation users actually want versus what might feel intrusive.
User-Centred AI Integration
I explored how AI can personalise recommendations based on both physiological data and inferred mood, while ensuring transparency in how suggestions are generated.
I also learned to communicate AI-driven features in ways that build trust with the user.