Virtual machine scale sets now provide the capability to automatically repair unhealthy instances based on application health status. Configure the scale set instances to emit application health by using either the application health extension or Azure Load Balancer health probes. After the automatic repairs policy is enabled, when an instance is found to be unhealthy, the scale set will automatically delete the unhealthy instance and create a new one to replace it.
To learn more about this feature and enable it for your scale sets, read the documentation.