Design Journey

Follow along our journey as we Discover, Define, Develop and Deliver!

Discover

To gain greater insight on the problems our solution should tackle, we conducted a series of field trips down to Apex Harmony Lodge in order to learn more about our target audience and the problems they were facing.

Define

User Personas

From our research, we identified our primary users: Senior and Junior nurses at Apex Harmony Lodge. From there, we created 2 user personas to better express the problems we identified in the scenarios.

Problem Statement

We also identified 3 main problems:

Language Barriers

Many residents communicate in dialects (e.g., Cantonese, Hokkien) or slurred speech that foreign nurses struggle to understand, leading to misinterpretations and delayed responses.

Information Overload During Onboarding

New nurses are quickly presented with large volumes of resident-specific information, often in unstructured formats. Without a systematic support tool, much of this knowledge is forgotten or inconsistently applied.

Limited Access to Assistance

Understaffed situations mean that junior nurses cannot always consult senior staff in time. This results in heightened stress for newcomers and interruptions to care continuity for residents.

Which led us to frame our final problem statement:

HMW leverage AI to create a service for healthcare staffs to streamline the process of communication between themselves and residents with dementia in Supported Living (SL) & Tender Loving Care (TLC) homes; in order to reduce the time taken for healthcare staffs to understand the unique traits and cues of different residents?

Scenarios

With the problems we identified, we developed a storyboard to illustrate the 3 problems faced by the nurses at Apex Harmony Lodge.

Develop

Systems Architecture

Tech Stacks

In developing our solution, we divided responsibilities by features and followed a strict timeline, with each team member taking ownership of a specific component. This approach allowed us to work in parallel, ensuring steady progress across multiple aspects of the service. As individual features were completed, we integrated them into the overall platform and refined the solution through collective reviews. Feedback from our professors and industry partner played a key role in helping us align our individual contributions into a cohesive and effective final product.

Design Iterations

Our design process was highly iterative, shaped by feasibility testing, ethical considerations and continuous feedback from stakeholders at Apex Harmony Lodge.

For more details, check out our Figma Prototype board.

First Iteration

Our initial concept involved a sensor-based system using microphones and cameras to capture auditory cues and visual micro-expressions, with AI classifying behaviours and alerting caregivers in real time. Although promising, user research raised key concerns—privacy issues around constant surveillance, the heavy technological demands of real-time processing, and the difficulty of accurately interpreting behaviours without context.

Second Iteration

We pivoted from surveillance to caregiver empowerment. ApexCARES now centers on tools that support staff: a Speech-to-Text Translator for communication, an AI Chatbot for guidance, and a Behavioural Database for preserving knowledge. This human-AI collaboration upholds our core insight—that subtle behavioural cues are key in dementia care—without relying on constant monitoring.

Final Iteration

ApexCARES is an AI-powered mobile and web platform that enhances communication and caregiving at Apex Harmony Lodge. It features real-time speech-to-text translation for dialects and slurred speech, a dynamic resident behaviour database that preserves vital care knowledge, and an AI Virtual Assistant (AVA) that provides contextual suggestions using institutional memory. Together, these tools help nurses overcome language barriers, adapt to new residents, and deliver more informed, dignified care.

Deliver

Final Solution: ApexCARES

ApexCARES is a mobile and web AI-powered platform that strengthens staff-resident understanding through three core features:

Slurred Speech-to-Text & Translation – Converts and translates dialects (Hokkien, Cantonese, Malay, Tamil) and slurred speech, helping nurses overcome language barriers in real time.
Resident Behaviour Database – PostgreSQL system storing medical histories, behavioural patterns, and caregiver notes, preserving institutional memory across shifts.
Apex Virtual Assistant (AVA) – RAG-powered chatbot that uses the database and LLMs to provide contextual care suggestions for new or unfamiliar behaviours.

Together, these tools turn subtle behavioural cues into actionable insights, enabling more responsive and dignified dementia care.

More details can be found in our Design Journey Report below!

SDS Final Report.pdf

Page updated

Google Sites

Report abuse