What is missing from HUMS is something that we wanted to have when it first started in the 1980s. A light in the cockpit which would say.
DO NOT FLY THIS HELICOPTER
or
LAND AS SOON AS POSSIBLE
The problem is that there is a huge gulf in certification between software taking measurements in flight that will be downloaded on the ground for analysis by engineers/computers to make an "it would be a good idea to manually check this bit of the aircraft" decision, and software that takes measurements in the air to present an indication to the pilot as to whether to continue flight or not. A huge gulf.
HUMS is predicated on the situation that there is sufficient time between a degradation becoming detectable and the subsequent failure of a component, that download of the data and ground analysis of it will detect the former before the latter. Trying to put general-purpose measurement and analysis functions onto the aircraft such that the result is instantly available to the pilot as a 'land NOW' indication is not only expensive, but it is hard. There are numerous false positives that could, conceivably, cause aircraft (and people) to be lost due to Land Immediately alerts that are wrong.
Given the general assumption within the use-case of HUMS, it is not actually a good vehicle for flight deck alerts - it is an indicator to an engineer that there could be a problem with a particular part of the aircraft and hence that should be checked. It is up to those checks to determine whether the aircraft should fly or not.
If there is a specific failure case where the time between a degradation being measurable and the subsequent failure occurring is too short for the 'standard' HUMS turnaround then there is a very good case for developing and installing systems to measure and alert on that specific situation. These would be developed to the appropriate certification level and the results (hopefully) be trustworthy enough to present on the flight deck.