Not So Smartphones
Many variables that were included in the smartphone study weren’t reported, and there’s a good reason for that: they were non-significant. I don’t mean not statistically significant—I mean not significant in the real world either. In fact, I don’t think I’ve ever seen a study with so many variables that were non-significant.
The one that surprised me the most was that simple tracking of blood pressure or the use of the artificial intelligence (AI) application did not change the percentage of people who took their BP medications regularly. Neither approach shifted the scale one little bit. I would have expected that at least some people would have started taking their meds regularly, but they didn’t in either group. In next Tuesday’s Memo, I’m going to give you a good reason to take your meds based on a recent COVID-19 study.
Many of the variables were from questionnaires, and as I often remind you, self-reporting is always suspect. But it raised a few questions about the reported results.
Questions
The first question I had was related to the statistical analysis they ran. Instead of looking for differences within groups, they ran comparisons only between groups. I’m not saying that any differences were profound, but it did appear that there were some that were interesting.
For example, the control group demonstrated no differences in servings of processed meats per week, but the app group decreased servings by about one-half portion. The servings of sugar-sweetened beverages decreased by about a half-portion per week as well, while there were no changes in the control group. That would seem to be a benefit and if a statistical analysis were run, it might have been statistically significant—take a win wherever you can get it. You shouldn’t overstate the findings, but it does support the idea that reminders about a healthier diet might be effective.
I don’t understand why the researchers used an application that was being beta-tested for use in the study. If the number of subjects was limited, and AI requires a lot more data points to really “learn” enough to decide what the subjects need to know and how best to present it to them, it seems the study was destined to fail before it began. If they had called it a pilot study to gain insight to propose a major clinical trial, that would be more logical because that’s what pilot studies are for: to decide whether larger studies are warranted. They came to the same conclusion, but focused on what they didn’t show instead of what they did.
The Bottom Line
The use of smartphones, tablets, and laptops together with applications designed to monitor health are growing in use. Paula recently had a consultation with a specialist and her first telemed physical. I think there’s a place for these types of electronic services, especially during this unusual time. But no matter how many subjects are used to train AI, I think it will always stop short of what they hope applications will do, because there’s no app that will get people to do what they don’t want to do. No logic. No mini-goal setting. No reasoning.
I always thought that education was the key. It isn’t. Even with my education, I have trouble doing the things that I know I should do for my health. The willingness to change has to come from within. Until people have that, no program, person, or application will help them achieve their health goals. It’s wrapped up in the third word in my tagline: what are YOU prepared to do today? It is and always be your choice.
Because we’ve talked about blood pressure all week and more is coming next week, this seems like an obvious time to offer you my High Blood Pressure download at half price; only a few CDs are left, also half price, and when they’re gone, they’re gone. Members and Insiders who log in first will get their discounts as well.
What are you prepared to do today?
Dr. Chet
References:
1. https://bit.ly/39OmUCc.
2. JAMA Open. doi:10.1001/jamanetworkopen.2020.0255.
