Memory statistics of each node

Any chance memory statistics of each node can be enabled in cloudwatch? I am currently in the process of scripting autoscaling and it would preferred to use memory metrics to trigger the scaling rather than the other options (cpu, disk io, network)