Challenges of a Medical Coding Software
Medical Coding is the process of finding a match of the verbatim to a term from the dictionary. It sounds simple, but the process is anything but simple. Experts would agree that medical coding is a science in itself. Understandably, to develop a software that performs medical coding, would be an arduous task. So what would be the challenges of a medical coding software?
- Finding a match: Finding a match of the verbatim to the term in dictionary is the most challenging task that a medical coding software needs to perform. There is a difference between "finding a match" and "finding the match", for example the software could find a match of the verbatim "headache" to the dictionary term "headache" with ease but may not find a match of the verbatim "fevr" to any dictionary term. A medical coding software should be intelligent enough to understand that if the verbatim is "fevr" it most probably means "fever" and then find a match for "fever" in the dictionary. The software could implement many kinds of fuzzy matching or approximate string matching algorithm to find a match, but that would not be enough. The challenge for the software is to determine the leeway to be given for fuzzy matches. Taking the example of headache, the match of verbatim "headache" to the dictionary term "headache" is easy as it is an exact match. But what if the verbatim is "head ache". The challenge faced by the software is to find out if the verbatim most probably means "headache" or "head pain" and then find a match in the dictionary accordingly. The real challenge is for the software when the verbatim and the dictionary term do not sound or look similar at all, and no manner of fuzzy or approximate matching can find any similarity between the two, for example the verbatim is "Retroauricular pain" and the match to a preferred term is "headache". If a medical coding software does that, it is safe to say that it has taken the challenge head on and emerged victorious.
- Quality of match: The challenges of quality of match goes hand in hand with the challenges of finding a match. The software may be able to find a match, but is it a good quality match? Finding a match for "Retroauricular pain" to preferred term "headache" should be considered a good quality match. The quality of medical coding depends on the quality of match of the verbatim to the term in the coding. A bad quality match will lead to bad quality of medical coding, which will have negative repercussions through out. It is imperative for the medical coding software to provide a good quality match.
- Performance: Performance is another challenge faced by medical coding software. If it provides good quality matches, obviously it is performing a plethora of calculations and logical operations which consume a good amount of CPU and thereby deteriorating performance. On the other hand if the software focuses on performance, it may let go of some logical operations it deems as lower in priority or unimportant and that leads to bad quality of coding. To be able to provide a good quality of coding and at the same time not compromising on performance is a challenge. To over come this challenge requires an in depth knowledge of just not the requirements of the medical coding industry but the technology in use as well. Shaving off 50 millisecond in the process of finding a match may seem trivial or unnecessary but when compounded with 500,000 verbatim to be coded overnight, it amounts to a time saving of close to 7 hours.
- Support for Different dictionaries: No two dictionary is alike. The challenge here is to be able to support any and all format of the dictionary. Supporting MedDRA and WHODrug is easy and pretty much granted, but to be able to support all known standard dictionaries and even custom dictionaries is a challenge which many medical coding software shies away from. The challenge is to provide one platform for all your coding needs across all dictionary.
- Integration: Many a times medical coding software need to integrate with other systems. To be able to support integration with any other system, is a challenging task and on top of that to be able to integrate any other system with the medical coding software not within the premise but hosted elsewhere is icing on the cake.
In conclusion not all coding software are alike. Its a prerogative of users and decision makers to determine if the coding software has taken the challenges head on and emerged victorious.
If you need to consult with me on any aspect of what I just said, please do not hesitate to contact me.
Punit, really interesting!
It is truly amazing how complicated such a seemingly simple task is. This is further compounded by custom dictionaries and proprietary workflows that many companies insist on maintaining. One example is a customer we worked with that was using a customer dictionary, based off of MedDRA, but containing over 1.2 million terms! Basically, rather than allowing search algorithms to find a proper match to the standard dictionary, they only allowed 1:1 exact matches from their "dictionary" This created for them a massive dictionary to maintain, plus created performance issues with the technology because of the sheer number of terms to be matched against. The biggest tip I would have for any coding group would be to look at how to optimize your coding workflow to take advantage of modern technologies.