Blog

Healthcare Summer Reading: Invisible Women

As artificial intelligence and machine-learning are increasingly touted as the solution for everything from shopping to healthcare, the need to better understand the data that goes into these solutions increases. We’ve written before about how trying to solve patient problems using only the EMR data delivers only half the equation because it leaves out everything that happened to the patient outside the clinic, and the patient’s own experience.

Machine Learning in Healthcare: How To Avoid GIGO

Self-Driving Healthcare

However, it turns out that this data may only deliver ¼ or less of the solution when it comes to women’s health, as the fact-packed “Invisible Women: Data Bias In a World Designed For Men” points out.

While Invisible Women tackles politics, healthcare, manufacturing, career, and finances, so many of the problem areas where key decisions are made without enough data result in population health and healthcare problems, even decisions about automobile manufacturing, snow shoveling, and portable cookstoves.

In example after example from every industry, the “normal” male is used as the standard, and women are seen as aberrations, resulting in health, safety, and finance inequality. One size fits all is actually one size fits average male.

If you need to make a case, either business or healthcare related, for the need for diversity of thinking, and for having the data for evidence-based decision making, this book will help you. You’ll also realize that in evidence-based decision making in healthcare, the data is missing for 51% of the population.

Examples include:

  • Crash test dummies that don’t approximate women’s bodies so that women are 17% more likely to die in a car crash.
  • Drug testing that does not require evaluation of outcomes by gender. (The UK does not require any gender evaluation for randomized control trials so researchers are advised to look at studies from other countries to ensure gender inclusion.)
  • Drug dosages that are not adjusted for size or hormones
  • Health trackers that underestimate women’s activities and don’t include menstruation tracking
  • Increased risk of hip fracture by making female solders match an arbitrary male gate length
  • Public transportation safety issues
  • Greater risk of women being misdiagnosed for heart failure because symptoms present differently
  • Portable cookstoves intended to decrease indoor pollution but aren’t used because they need constant tending, and mean that women can’t get other chores done.

The list goes on. Practically every paragraph in this book has a practical example where getting the right data, either qualitative or quantitative would have resulted in better quality of life for women (and everyone really.)

The good news is that this implicit bias that normal can be overcome with a contentious approach to collecting data and feedback, and a rigor of examining the data and outcomes by gender. Interestingly, when reading this book, I realized that this should be another way that we evaluate our Wellpepper care plans. We currently mostly segment our data analysis by age because there has been previous skepticism about older patients ability to use technology. Now I’m thinking that gender differences in care plan outcomes might be a really interesting source of insight. What might we learn about recovery? We know women experience pain and medication differently. Are their gender based clinical insights in our outcome data as well?

While Invisible Women is probably not beach reading, it’s still highly recommended book to add to your healthcare and data reading list.

Posted in: big data, Clinical Research, Healthcare Research, Healthcare transformation, machine learning, patient-generated data

Leave a Comment (0) ↓

Leave a Comment

Google+