I find humor in you continually stating that MTM can be done by computers. While this is absolutely true, it's not the only thing that computers can do, cheaper, faster and more accurate. Leaving the entire human/computer interface out of the equation, you do realize that a computer can do any repetitive, structured set of events, including differential diagnosis.
Of course I considered this even before I went into medical school. Can the physician be replaced by computers? I concluded that the answer for the foreseeable future is no. Here's why. For something as simple as bronchiolitis, will the computer be able to interpret the lung sounds? Will it be able to interpret the breathing pattern? Nope. Someone has to input that data into a program. Maybe that person is a physician, PA, NP, RN, MA, etc. The inputting of the data is the key. A program is only as good as its data and algorithms. The more experience and trained a person the better the input. Inputting data is trickier than you may think. Right now, most data capture is just a drop down box. What if the choice you want isn't there? Computer programs have a really hard time understanding if you just free text things in because they can't do comparisons and look things up well. What if the input is wrong because the person doesn't know what the difference between a wheeze and a rhonchi is? If the data going in is crap, the output is crap. That's why a program can at best spit back a billion ddx's and give what it thinks is most likely, but the clinician has to determine what it really is. There's a reason why fields such as IT, accounting, finance, etc can be outsourced while the neighborhood plumber doesn't have to worry (except for illegal immigrants but that's another topic).
How is MTM different? Because the data would already be in a database and it's well-defined data. "Aspirin 81 mg PO BID" There's no haziness about it. If you know what the data will look like, you can write rules to determine drug interactions, potential side effects, best dosing, etc. The key to developing any application like this is always the data. If you trust the data and you know the rules (which you can get by hiring an army of pharms), you can build a program.
So what would such a system look like? I would feed information about allergies, past and current prescriptions, previous drug reactions, etc into it and it would spit back medication recommendations based on MTM or whatever will be the gold standard. Since I would know the patient and his medical history well, I would be able to accept or reject recommendations. This could all be done in seconds and there could be multiple iterations until we're satisfied with the recommendations. Think of it as the "UpToDate" of pharm. For an analogy that you may know, think of TurboTax software for doing your taxes.
I just hope somebody doesn't beat me to this idea, but somebody probably will.
🙁 Seriously, if someone sees that pharms can charge $140/hour for doing MTM x the number of pharms doing this nationwide, you start to see why somebody or some company would develop such a system. If I can see this, you can bet people who have more time and funds than I do see this too. The people in Silicon Valley drool over this stuff.
