Thoughts on Mental Health Tech

I recently read this article on a potential California system to identify and push along services to the mentally ill in order to better support and service their journey to healthy. Frankly, I don’t think people realize how useful it can be (as long as the service is very much opt-in).

The most important priority in any person with a mental illness and/or serious mental illness is to get healthy. The disease they have affects behavior based on internal mechanisms of something that’s, right now, unnoticeable. Generally, what that leads to is a lack of understanding and help for the individual to go anywhere in life until they’ve shown and proven to be healthy and productive to any new person. So, the question arises: how do you know someone’s healthy with a mental health condition? What are the factors to look at? Are they a “danger” to themselves or anybody else?

Doctors, as always, can give an approval that someone is healthy enough to contribute back to society (though that’s a tough situation to be in to give approval about someone’s human rights to the pursuit of life, liberty, and happiness). Also, some peer and support network can give the a-ok that they’re ready to go (in the best case scenario). Though, no one really has a standardized plan for these efforts, as of yet, on any real scale. Also, unfortunately, medical science just can’t pick up on real-time internal mechanisms for behavior right now (at least I haven’t seen anything great and useful).

The technology that is available, like the Apple Watch and Fitbit, are trying to get there alongside other apps that track your CNS, but it’s really not accurate enough. Plus, they’re all new, unproven tech that is going through growth and needs to be researched and proven in clinical trials, IMO. Most monitors right now, like the fMRI scan and blood tests, are static.

I’ll disclose: Personally, as a person with a 10+ year tech career and a good understanding of how some of this stuff works…as well as with a serious mental illness myself (yes, very much stabilized now), I want to prove I’m capable of leading a very productive life to anyone new I meet. That’s where, I think, Mindstrong has a potentially fully realizable dream. It’s where the 20% of the population suffering from a mental illness can fully say, “yes, I’m either unhealthy or healthy enough to contribute to society on any level and to any person or group of people”. Incredibly important and a hugely untapped market.

I don’t give one crap about the privacy concerns, as long as I opt-in, because I’d gladly give that up, in my situation, in order to live a great and productive life again (the same or better one than I had before – which is saying a lot). That’s the state I and many others are in from what I’ve seen. Not a good situation to be in life. Further, a lot of this tech would help drastically improve scientific understanding of a very unknown set of illnesses. That’s worth it for any set of privacy concerns to get and gain stability.

Climate Mood

Continuing with the delve into Climate Change, looked into understanding how people feel, positive/neutral/negative, about the subject on Twitter. Using TextBlob, with Tweepy, Pandas, and Plotly Dash…I built initial revision of a data visualization dashboard, Climate Mood, that analyzes tweets over a sample of the last 2,000 collected hourly. The idea is to provide a simple, sentiment-focused look across multiple analyses. I’ve particularly enjoyed building and re-applying some of the methodologies and statistical analyses I’ve learned recently.

Go to Dashboard

As mentioned, the dashboard pulls from Twitter over an hourly basis and deposits into a PostgreSQL database. The top chart provides a linear regression trend line (unevenly for now) across tweets collected that can be toggled. Also, tweets can be seen on the top-right tweet-box on hover. Other simple sentiment analyses include average verbosity of tweet length, popularity of tweets, and average sentiment-based amplitude of tweets. Feel free to check it out!

Looking forward to applying more statistical analyses across these auto-updated samples. If you have any feedback or suggestions, please submit an issue on the GitHub repo or contact me.

Climate Trends

Been thinking about climate a bunch lately and started looking into trying out the latest data tools and trends out there on the Internets. I’ve particularly enjoyed R and Pandas/Python, especially via Plotly Dash, and re-applying some of the methodologies and statistical analyses I’ve learned while on the job. The latest curiosity is in global temperature change since 1910 and various other societal factors. Lo and behold, the first iteration and not-so-polished, Climate Trends.

You can view the dashboard here

The dashboard has a drop-down menu of various analyses around global temperature change and linear fits across birth rates, fertility rates, death rates, life expectancy at birth, and rural population growth rate as well as a nominal years passed chart. Feel free to dabble!

If you have any feedback or suggestions, please submit an issue on the GitHub repo, respond on Dribbble, or contact me.

Note on Product Prioritization Frameworks

Always looking for ways to improve how to prioritize features appropriately to a roadmap. Each sprint is full of items to juggle between buckets of customer requests, items that improve customer retention, or (of course) increasing revenue. Features working towards on the roadmap are no doubt important, but fixing an issue customers see out in the field could potentially improve retention and apply a greater impact to the company’s bottom line.

The 1st approach calculated “ROI” by dividing using an averaged set of “Impact” ratings using a simple 1-5 likert scale (by my Eng Lead, UX Designer, Technical Support SME, and myself) and then divided the average by “Cost” (i.e. t-shirt sizing) to find ROI. It worked reasonably well; we were able to get to a prioritized list of features, but there was a skewing towards items that were quick wins and less towards major user features.

So, taking in some of the great PM content out on in the interwebs – I took the cue for my own team by trying to find a methodology. The latest approach uses weighted-values for 2 top-level product themes and “customer satisfaction” to identifying priority for each backlog item. Thus far, the priorities for each backlog item have all passed my PM gut. Also, there’s been solid feedback from the team too.

Take a look, hope you find it useful! If you find a way to improve on it, would love to hear about it.