A neural network defines health as…

One of Mike Lynch’s bot creation is the Glossatory, which generates definitions of words based a recurrent neural network. Mike illustrates some of the more outlandish ones at @glossatory@weirder.earth

a neural net defines AUTHORITY, created by Glossatory/Mike Lynch 2019. Shows a drone-like machine looking over a rough sketch of a village.
a neural net defines AUTHORITY, created by Glossatory/Mike Lynch 2019. Source

The Glossatory is based on a recurrent neural network, which is the type of artificial neural network that’s typically used for machine translation. Mike used a training set of 82,115 definitions that were taken from WordNet. The results are often amusing, frequently incomprehensible, but mostly they’re just odd.

Continue reading “A neural network defines health as…”

Improving public health messaging

I wrote this post for Croakey and Reporting on Health a little while ago. The ideas are relevant to a lot of health-related communication.

If you follow health reporting on TV or in newspapers you could be forgiven for thinking that the only things that happen are scandals in clinical services or trials of new drugs.

But journalists and PR people aren’t necessarily all to blame. As health professionals we often do a pretty bad job at explaining what we’re doing.

I’ve been thinking about how we can do a better job of explaining the slightly more complex interventions that are required for a lot of the population health problems we face. New drugs and clinical stuff-ups fit into well understood tropes. We don’t have to explain everything, the audience can take shortcuts because they understand what type of story it is.

When it comes to population health issues it’s often not as easy because the issues are interdependent, and many of the interventions are unfamiliar to a mainstream audience.

A lot of my work is on health impact assessment, which involves developing evidence-informed recommendations to inform decision-making and implementation.

When it’s explained like that, it’s no wonder journalists aren’t interested. It sounds like a technocratic snooze-fest.

Instead, let’s think about the demand for new housing in most Australian cities and the pressures to release new land on the fringe. Also think about how disastrous the design of some new suburbs has been for population health in the past by promoting car dependence, limiting walkability and increasing social isolation. The design of our suburbs matters.

Health impact assessment has been a practical way to get people to think about the health consequences of the way suburbs are designed and here’s a few examples…

Contrast that story with the bland description of health impact assessment in the earlier paragraph. It sounds a lot more engaging.

We often fail to describe this broader story in population health, not just when pitching stories but also when we communicate with other sectors (or even within the health sector).

Here are four ways I think we could get better at messaging.

1. Don’t explain the solution, explain the problem

Or better yet, explain the causes of the problem. This piece from the Atlantic Cities is a good example. It describes the phenomenon of “ghost estates” in Ireland, which came about when 2,800 housing developments were abandoned as a result of the GFC. A community group has started planting trees on these sites to reintroduce some aspects of nature into these abandoned building sites. The piece works because rather than leaping into a description of the NAMA to Nature group, it first describes the problem as well as its causes. Too often we forget the broader context when describing what we’re doing in population health.

 2. Don’t rely on the usual suspects

There’s evidence that people are more willing to listen to arguments when they come from unexpected sources, at argued in this Ramp Up post. A good example recently is the conservative economist Judith Sloane’s calls for an increase in the NewStart allowance. She could hardly be described as a usual suspect when discussing the rights of the unemployed. Different people paid attention to her comments as a result.

3. Avoid jargon and language that alienates people

The importance of this is emphasised in the Robert Wood Johnson Foundation report A New Way to Talk About the Social Determinants of Health. People switch off when things are described in stereotypical or politicised terms. As health professionals we often tend to fixate on the solutions and the jargon that surround them, partly because it’s what occupies most of our time but also because we are already convinced about the importance of the problem.

4. Describe the human impact

This is often the hardest part for population health stories. Individual stories can illustrate broader population issues but they can also be misleading and seem glib. We often understandably resist this because we have an obligation to respect the dignity and privacy of the people we work with and we can’t control how their experiences will be reported. Without the human dimension though it’s difficult for not only journalists but also the audience to connect with the story. Human-scale narratives still matter, even with the most abstract ideas.

A good example of a media piece that embraces history, messiness, complexity and a population approach is Melissa Sweet’s description of Miller for Inside Story. It’s a story about the problems faced by a suburb in South West Sydney.

It’s an almost impossible story to convey in usual journalistic form because it has a lot of history and people involved and doesn’t have a neat narrative arc or resolution. Despite this, the piece manages to convey a lot of the complexity to the reader and provides a number of insights. The point is that it is still possible to tell even the most complicated stories in engaging ways.

We’re confronting big social and population health challenges but we have some ideas about the solutions. We just need to make sure we don’t bury the lede.