What challenges did Chulalongkorn University face before AI implementation?

Scale mental health pre-screening and hotline triage during high call volumes. Reduce suicide risk and improve response times while protecting sensitive user data. Maintain reliability and compliance while handling peak demand.

What was the impact of the AI solution?

The DMIND app screened over 300,000 users and identified over 7,000 high-risk individuals for timely intervention. The voicebot processed 100,000 calls within six months, with 40 percent of callers adopting the system. Severe-case response time improved from 30 minutes to 7 minutes. The system maintained 99.9 percent availability. AIMET avoided the need to hire three to five additional staff compared with an on-premises setup.

Rank#1,580of 3,449

Source

ProductionEvidence: Medium65/100

AIMET Case Study | Amazon Web Services — AI-powered mental health screening and voicebot on AWS

Use case typeVoice automationUpdated Nov 1, 2024

Chulalongkorn University’s Center of Excellence in Digital and AI for Mental Health (AIMET) needed to scale mental health pre-screening and hotline triage during high call volumes, reducing suicide risk and improving response times while protecting sensitive user data. The solution used AWS to run DMIND, an AI-powered mental health voice/text pre-screening app, and an AI voicebot for Thailand’s national mental health hotline. The implementation used secure, scalable AWS infrastructure with AWS Fargate, Amazon Elastic Container Service, Amazon Route 53, Elastic Load Balancing, Amazon ElastiCache for Redis, Amazon VPC, and Amazon SQS. AIMET also used a web application firewall and CDN services to protect data and deliver low-latency access. The architecture dynamically scaled during peaks, routed critical cases, and supported both pre-screening and hotline triage workflows.

Organization: Chulalongkorn University
Industry: Healthcare
Location: Thailand
Published: November 2024

Reported outcomes

300,000 users

users screenedAdoption & scale

7,000 usershigh-risk individuals identified100,000 callscalls processed40%caller adoption76.7%severe-case response time reduction99.9%availability

Strategic outcomes

Risk & complianceProtected sensitive mental-health data in a secure cloud environmentSpeed & agilityEnabled timely intervention for high-risk casesScale & capacityExpanded mental health support to underserved communities

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Users screened: 300,000 users

AWSNov 1, 2024Customer storyExplicit claimMedium evidence strength

screened over 300,000 users

Normalized claim

High-risk individuals identified: 7,000 users

AWSNov 1, 2024Customer storyExplicit claimMedium evidence strength

identified over 7,000 high-risk individuals

Normalized claim

Calls processed: 100,000 calls

AWSNov 1, 2024Customer storyExplicit claimMedium evidence strength

processed 100,000 calls

Normalized claim

Caller adoption: 40%

AWSNov 1, 2024Customer storyExplicit claimMedium evidence strength

with 40 percent of callers adopting the system

Normalized claim

Severe-case response time reduction: 76.7%

AWSNov 1, 2024Customer storyInferred claimMedium evidence strength

reducing response times for severe cases from 30 minutes to just 7 minutes

Normalized claim

Availability: 99.9%

AWSNov 1, 2024Customer storyExplicit claimMedium evidence strength

maintained 99.9 percent availability

Normalized claim

Staffing avoided: 3 staff

AWSNov 1, 2024Customer storyExplicit claimMedium evidence strength

avoid the need to hire an additional three to five staff compared to an on-premises setup

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Chulalongkorn University, Center of Excellence in Digital and AI for Mental Health (AIMET)
Provider: AWS
Maturity: Production
Linked source: AWS

Deployed an AI voicebot for the national mental health hotline to prioritize urgent cases and support call handling

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 2 of 2

1Voice automation
2Customer service automation

Scale mental health pre-screening and hotline triage during high call volumes.
Reduce suicide risk and improve response times while protecting sensitive user data.
Maintain reliability and compliance while handling peak demand.

Built the DMIND AI pre-screening app to assess mental health via voice and text and categorize users into four severity levels.
Deployed an AI voicebot for the national mental health hotline to prioritize urgent cases and support call handling.
Used AWS Fargate and Amazon ECS for dynamic scaling, Amazon Route 53 and Elastic Load Balancing for traffic routing, Amazon ElastiCache for Redis for low-latency performance, Amazon VPC and a web application firewall for security, and Amazon SQS for reliable microservice communication.

Technologies

AWS Fargate Amazon Elastic Container Service Amazon Route 53 Elastic Load Balancing Amazon ElastiCache for Redis Amazon Virtual Private Cloud Amazon Simple Queue Service

The DMIND app screened over 300,000 users and identified over 7,000 high-risk individuals for timely intervention.
The voicebot processed 100,000 calls within six months, with 40 percent of callers adopting the system.
Severe-case response time improved from 30 minutes to 7 minutes.
The system maintained 99.9 percent availability.
AIMET avoided the need to hire three to five additional staff compared with an on-premises setup.

Architecture

AWS-hosted scalable mental-health screening and hotline triage system using serverless/containerized compute, load balancing, caching, queueing, VPC isolation, and web application firewall protection.

Sources & evidence1

Evidence: Medium65/100Evidence strength