Complying with medical prescriptions

Patients are not very good at adhering to their medical prescriptions. Here’s how a little cloud computing and mobile technology can help.

The medical community refers to Adherence as the degree to which a patient correctly follows medical advice, for example completing a prescription to treat an acute cough.   Not surprisingly, patients are not very good at adhering to their  prescriptions, there are many reasons why.

In some cases, information technologies can help improve adherence.  In this post, I use Microsoft Azure’s Logic Apps service   to automatically convert my family’s medical prescriptions into timely calendar  entries that popup as reminders on my iPhone. 

First, a brief summary for new visitors. A few years ago I started tracking key aspects of my family’s day-to-day well-being.  You can browse the full list of variables I am tracking here.  I use this data to conduct casual exploration and N-of-1 experimentation to address issues and opportunities affecting my family’s growth, health and happiness.

reminder.png
Auto-generated prescription reminder in Google Calendar
logic.png
Microsoft Azure Logic App to convert medical prescriptions into mobile phone reminders

 

Bouncing back from stress

Day-to-day demands and pressures sometimes get the best of us.  Whether reacting to new assignments, unplanned issues at work, or juggling after school activities for the kids, sometimes it feels these demands accumulate to a point that exceeds our ability to handle them.  If we begin to worry at this stage, stress sets in, making it more difficult to regain the focus to move forward in the best way possible.

In every waking hour we are being triggered by demands, in the form of people, events, and circumstances that have the potential to change us.[2] Sometimes the demands accumulate towards a tipping point, other times it takes just one to set the ‘worry’ chain in action.  Although we cannot control the influx of these demands in our lives, we can improve how to respond to them.

Over the past three years, I have been tracking occasional periods of stress in my life.  I am learning more and more about what triggers them and how best to cope.

My goal is to strengthen resilience and minimize or completely eliminate this occasional stress.   This n-of-1 experiment tests the effectiveness of my resilience-boosting activities in reaching this goal.

stresscount.png
Figure 1: Days with stress 2016 – 2018

Keywords
stress, resilience,  n-of-1, quantified-self

Glossary
Pressure –  a “demand to perform.” The demand might be intense, but there is no stress inherent in it, and as we’ll see, the key to resilience is not to turn pressure into stress.[1]

Resilience – the ability to negotiate the rapids of life without becoming stressed.[1]

Reflection – the process of thinking over a problem to arrive at a solution.  What is missing from reflection is catastrophizing.[1]

Rumination – worry or the constant churning over what-ifs and if-onlys. Its what causes stress.[1]

Stress – pressure + rumination.[1]

How did I do it?
My self-tracking experiments are powered by ostlog – an open-source Personal Well-being Library. Since 2013, I have been using ostlog to track a broad set of variables covering spiritual, social, physical, emotional, intellectual, financial and environmental aspects of my family’s well-being. I use this data to conduct casual n-of-1 experiments such as this one.

The first step is to quantify just how much stress (see glossary) I was experiencing over the past couple of years. Figure 2 shows the number of days (per month since 2016) where I experienced stress. The chart confirms a downward trend I suspected, with peaks of six days in March and October 2016 and down to a total of three days during the first four months of 2018.

sd.png
Figure 2: Stress days 2016 – 2018

The chart also confirms a general improvement in navigating the planned and unplanned demands of day-to-day life without becoming stressed by them.

What did I learn?
A few years ago I read Marshal Goldsmith‘s “Triggers: Sparking Positive Change and Making it Last”.[2] The book gave me a better understanding of positive change, motivation and difference between active and passive improvement practices.

Throughout my adult life, i’ve known and preached the mind and body benefits of regular exercise. With the arrival of our first daughter in 2010/2011, my exercise routine went dormant until the following year (see figure 3.) This general pattern repeated itself in the subsequent years, as we welcomed new additions and responsibilities in the growing family.

ed.png
Figure 3: Yearly exercises

I exercise in order to stay balanced, healthy and strong.  These benefits are hard to come by when when inconsistency creeps into my exercise routine. As Goldsmith points out, “inconsistency is fatal for change.” In my case, my exercising remained pretty inconsistent until mid 2017.

The same can be said for learning opportunities.  Here I am referring to reading and writing I do during leisure time (see figure 4.)   Whether reading a great book or writing a blog post like this one,  learning provides some benefits similar to those associated with consistent exercise.   Curiously, figure 4 shows an increase in learning opportunities that is similar to my exercises during the same time period.

lo
Figure 4: Yearly learning opportunities

In their book, “Work Without Stress: Building a Resilient Mindset”,  Derek Roger and Nick Petrie emphasize the concept of resilience, and its role in helping individuals avoid stress.   Resilience, they say, is a skill that can be acquired by training and practice.[1]

In my case, I believe consistent learning and exercise are two ways that build and strengthen my resilience.  This experiment is really about using data to prove this.

Just to recap what I have shown thus far.  Figure 1 and 2 show a decrease in stress since 2016.   Figure 3 and 4 instead show an increase in exercise and learning during the same period.   Are the two linked?  Can the combination of consistent learning and exercise make me more resilient to stress? Will more of both help me reach my goal?

Results
So now I need data to support the belief that exercise and learning help me become more resilient.  The data in figure 5 shows a correlation between stress (y axis) and resilience building activities (i.e. exercise + learning) aggregated monthly during the period of interest.    The results are not conclusive (p-value was quite large), and more data points are needed to definitively prove this.  I am also not accounting for the many other other factors, including nutrition, family time, sleep quality, social activities and more, that help strengthen our resilience and avoid stress.

correl
Figure 5: Correlating stress (y-axis) with resilience building activities (exercise and learning on x-axis)

Check back for future updates as I collect more data points to test whether exercise + learning strengthen my resilience enough to help minimize or eliminate occasional stress.

[1] Work Without Stress: Building a Resilient Mindset (link)
[2] Triggers: Sparking Positive Change and Making it Last (link)

DIY Growth Charts

Today I’ll discuss do-it-yourself charts I use to track growth for my daughters.

First, a brief summary for new visitors. A few years ago I started tracking key aspects of my family’s day-to-day well-being.[1] You can browse the full list of variables I am tracking here.  I use this data to conduct casual exploration and N-of-1 experimentation to address issues and opportunities affecting my family’s growth, health and happiness.

I use Microsoft’s excellent PowerBI data visualization tool to mine through the data and create visualizations that make sense to me and my family.

For example, see these growth charts comparing height and weight for my three daughters to CDC growth standards (see figure 1 and 2.)

weight
figure 1: comparing daughter’s weight to CDC Growth Chart averages

Visual Design

I’ve added a few elements to make these charts easier to understand. First, I needed to plot the CDC Growth Chart averages as a reference point. These datasets contain the percentile averages for height, weight and other growth variables, for each month of growth, across a diverse sample of the population. In his data visualization book, Now You See It, Stephen Few talks about the role of pre-attentive attributes in preparing the user’s focus during the visualization. For these growth charts, I used color and shape (dotted lines) to highlight the range of percentile groupings and gently ‘push’ them to the background. In plotting the actual values for my daughters’ height and weight, I emphasized this line (solid red), allowing it to ‘call’ for our attention while also comparing values to the CDC averages. PowerBI supports the features to configure pre-attentive attributes this way. In addition, it allows me to easily create, publish and access these charts from the web, on my mobile phone and in a secure way.

height
figure 2: comparing daughters height to CDC Growth Chart averages

And this is quite helpful when providing context in general discussions with our pediatrician.

[1] My self-tracking projects are powered by ostlog – an open-source Personal Well-being Library

Personal vs. Household Analytics

A few years ago I started tracking key aspects of my family’s day-to-day well-being. You can browse the full list of variables I am tracking here.  I use this data to conduct casual exploration and N-of-1 experimentation to address issues and opportunities affecting my family’s growth, health and happiness.

The variables I collect represent the individual members of my family. For example, how many workouts did I do to the past month, or what is the average sleep duration for my daughter during Winter months. When reporting through dashboards and indicators, the variables are aggregated in time and space but the unit of reporting remains the individual.

worship
% of weeks in 2018 where family attended Sunday Mass

Recently I began experimenting with household-level reporting — the family as a single unit of reporting. For example, how many times did family attend Sunday Mass in the current year, or how many sick days did we experience last winter.

Aggregating by family this way made me realize a few things. First, there is a whole world of household well-being indicators waiting to be explored. Here I am not only referring to household well-being as an aggregation of the individual level. Rather, well-being indicators that only make sense at the household level. Second, reporting on well-being this way raises interesting new questions regarding the role of family in helping improve the well-being of its members.

Tracking sleep and meal duration

It’s been a while since my last post.  Here are some new things I’ve been doing and reading during this time.  There’s a lot to talk about!

On the data collection and analysis side of things, I started tracking new variables including mood, energy levels, sleep/meal duration and quality, continuous-learning activities, ongoing projects, and Church attendance.[1] This data joins twenty-three other aspects of my family’s well-being that I’ve been tracking since 2013.  Together the data is giving me a holistic view of our growth, health and happiness across the six dimensions of well-being I care about — Spiritual, Social, Intellectual, Physical, Financial and Environment.

For example, I now have indicators tracking my family’s sleep duration goals over the course of the year (see figure 1.)

sleep
figure 1 – tracking percentage of sleep lasting 7 hours for adults and 9 hours for kids over the course of the year

A separate indicator allows me to track time we spend at the dinner table.  The indicator was inspired by OECD and their chart showing the time spent eating & drinking each day across countries (see figure 2.) It also supports my belief in quality time together at the dinner table.

Eating-and-drinking-550
figure 2 – OECD Time spent eating and drinking

The indicator tracks the percentage of dinners lasting at least 40 minutes, with a goal of at least 80% each  year.

dinner
figure 3 – Percentage of family dinners lasting at least 40 minutes

These are just a few examples.   I am tracking fifteen more indicators and dozens of time-series charts across the six dimensions of well-being I mentioned earlier.  I am also working on a new experiment in the area of stress, anxiety and resilience. I hope to cover this and other aspects of family-tracking in the coming months – stay tuned!

Regarding general news and trends in the area of personal health tracking, digital well-being, personal informatics or, more generally the quantified self, it definitely feels that we are in the midst of a bubble of great promise and hype.  Three general observations I have during this period are:

  • The digital divide is real.  The market for digital well-being / personal health tracking will continue to be constrained by inadequate data skills in citizen scientists.
  • Collection fatigue is also real.  This applies to the commercial wearables industry.  In my opinion, these companies are only stressing out their customers; weighing them down with low-value data collection tasks.  Without a clear path to transform raw data into truly useful and actionable insights, customers will continue to be underwhelmed and dissapointed.
  • This creates a big opportunity to provide users with actionable well-being insights without overburdening them with data collection, or the need to acquire sophisticated data science skills to make sense of the data.

[1] My self-tracking projects are powered by ostlog – an open-source Personal Well-being Library.

“Alexa, how much do we pay the…”

A few months ago I wrote about the possibilities of integrating my family’s self-tracking data with smart home assistants, like the Amazon Alexa powered Echo Dot. One of my near-term self-tracking goals is to give members of my family the possibility of also deriving value from this data, and the Echo Dot represents one way to achieve this.

echodot

First, a brief overview for those of you new to this blog. A few years ago I started tracking key aspects of my family’s day-to-day well-being including symptoms, medicines taken, doctor’s visits, activity, vitals, finances and more. I use this data to conduct casual exploration and N-of-1 experimentation to address issues and opportunities affecting my family’s well-being—a process I refer to as Family Data Science.

My self-tracking projects are powered by ostlog – an open-source Personal Well-being Library.  ostlog works great for my needs, but not so much for my wife, who prefers simpler access to the information.  Today I rely on Microsoft PowerBI, generic SQL tools and other software for collecting and managing this data.   The Amazon Echo Dot opens the possibility of providing  natural language interface to query the same data.

We  own a single Echo Dot that sits on our piano in the living room. The kids use it to ask Alexa to play their favorite songs from their favorite films (e.g. “Alexa, play Trolls on Spotify”.). My wife uses it to stream her favorite radio station from Argentina (i.e. “Alexa, play radio maria”.)   As for me, the first step was to create a custom Alexa skill that responds to requests for well-being insights, queried directly from my self-tracking database.

Thanks to Amazon and Microsoft’s cloud serverless services (i.e. Lambda, Azure Functions), accomplishing this turned out to be a piece of cake.  This was the first Alexa request implemented:

Alexa, start ostlog

Alexa, how much do we need to pay the baby sitter this week?

With this new Alexa skill, my wife now has a hands-free way to access the self-tracking data, no special software required.   And with this personal finances related request, we no longer have to fumble through devices and software applications in order to retrieve this data (while the sitter waits patiently at the end of a long day and week!)