Implement HTTP health checks on the NLB with the application's URL.

Introduce a cron job on EC2 instances to scan logs for HTTP errors, triggering an application restart if found.

Switch to an Application Load Balancer, set up health checks with the application URL, and automate instance replacement for unhealthy ones via Auto Scaling.

Establish a CloudWatch alarm for the NLB's UnhealthyHostCount metric to trigger Auto Scaling actions for unhealthy instance replacement.