Can mathematical models predict and prevent the next suicide?

MMental health is, in some ways, the black sheep of public health: chronically underfunded, undervalued and neglected. One of the reasons has to do with the social stigma surrounding mental health. After all, it has long been considered awkward to seek treatment for issues like depression (while TB is a much sexier disease). An even bigger reason has to do with the complexity of deciding where to channel mental health resources – a process known as psychiatric epidemiology.

As computer engineers continue to produce increasingly powerful algorithms, models of complex systems have become the cornerstone of epidemiology. These programs collect massive amounts of data and combine it with a specific set of metrics to map public health issues, identifying geographic regions or demographics most at risk for a given disease.

Most epidemiological models have been limited to infectious diseases, predicting the spread of diseases like malaria and COVID. But some researchers believe it is time to apply these tools to mental health services as well. These scientists aim to create models to predict where problems such as severe depression and suicide are most likely to occur, and which interventions are most effective. By doing so, they hope to correct some of the mental health funding disparity.

The challenge: getting political decision-makers, psychiatrists and the general public on board.

Everything seems to be important for mental health, which makes studying difficult and fun.

— Daniel Eisenberg, University of Michigan

This is the question that occupies most of Jo-An Occhipinti’s time. As a data scientist at the University of Sydney in Australia, she was inspired after working on models to help control malaria outbreaks in the South West Pacific. Applying the same approach to mental health policy seemed like the next logical step. “I don’t think people realize how critical mental health is to the proper functioning of society,” Occhipinti told The Daily Beast.

However, for many, COVID-19 has pushed mental health back into the conversation. A recent World Health Organization survey found that the pandemic has triggered a 25% increase in depression and anxiety diagnoses worldwide. Along with this awareness comes a renewed interest in investing in mental health as public health. But it can be difficult to know where to start for public health facilities.

“Everything seems to matter to mental health, which makes studying difficult and fun,” Daniel Eisenberg, a mental health policy expert at the University of Michigan, told The Daily Beast. Socioeconomic status, brain chemistry, family ties, and general physical health are just a few factors that can put a person at greater or lesser risk for depression. For this reason, some epidemiologists and clinicians believe that mental health is simply too multifaceted to be studied using models.

However, Occhipinti does not view this complexity as a barrier to effective systems modeling, it simply means that models need to take these factors into consideration. To do this, Occhipinti and his team work closely with people who have first-hand experience of the problems they model. She recalls a case in 2017, when her team was working on a suicide risk model in hopes of developing a more supportive infrastructure for suicide prevention and recovery. There was just one problem: the model didn’t work. No matter how much data they fed, the results didn’t seem to match reality. “We just didn’t know what was missing,” recalls Occhipinti. “We thought we had everything.”

So they asked for the opinion of people with lived experience. Things clicked when someone pointed out that they lacked something called a “black hole,” a percentage of patients who receive immediate but inadequate care and bounce back from long-term treatment. Taking this variable into account, Occhipinti said, completely changed the model. Suddenly, the simulation and the real world aligned. “Once we put that missing piece in, the model just clicked into place. It was really quite amazing.

Of course, effective models only matter if you can use them to shape policy. “We could have these fancy models and so on, but we don’t want them sitting in an ivory tower,” Patricia Mabry, systems scientist at the non-profit HeathPartners Institute, told The Daily. beast. “We want them to be used.”

Mabry works closely with policymakers from the National Institute of Health at the Food and Drug Administration to develop public health guidelines based on detailed scientific models. This type of work is crucial, said health economist Steven Dehmer, because it allows scientists to project the impact of a health crisis without having to put the public at risk. “Especially when you have to make decisions in the absence of other data, it can really be helpful,” he told The Daily Beast.

Dehmer, who also works with the HealthPartners Institute, builds models based on a combination of clinical data and personal experience to help predict everything from the prevalence of lung cancer associated with smoking to the impact of high sodium in fast food on heart attack risk. Like mental health, these models rely heavily on sociological factors and individual behavior, and they can be refined based on ongoing inputs. Dehmer and Mabry see potential in using models to inform mental health policy as well, provided the intended recipients buy-in.

This process is as much a human process as it is a technical one. We need to get policy makers to see the value of this kind of work.

— Jo-An Occhipinti, University of Sydney

“It can be hard to convince a patient to make a change or do something they don’t believe in,” Mabry said. “You’re not going to be able to do it with just ‘Oh, trust me.’

A recent pilot program in Melbourne, Australia used Occhipinti’s dynamic models to identify people at high risk of death by suicide (participants had made at least one previous suicide attempt). The program offered 11 intervention options, ranging from removing potentially dangerous items from participants’ homes, to drug screening, to building community support networks. Not one of the 40 participants died during the trial. Building on this success, the Australian government has since invested an additional $2.3 billion in suicide prevention.

Even when good mental health services are available and public trust is high, some people may not take advantage of them. “A lot of people just don’t prioritize their mental health enough,” Eisenberg said. Others may not have enough free time or money to spend on mental health care, while still others may be deterred by stigma (real or imagined).

But by looking at people’s lived experience every step of the way, Mabry and Occhipinti hope that trust in public health initiatives and awareness of their importance will continue to grow. Currently, pilot programs similar to Melbourne’s are underway in the UK and US, but it remains to be seen whether they will result in more government funding.

“This process is as much a human process as it is a technical one,” Occhipinti said. “We need to get policymakers to see the value of this kind of work.”

Comments are closed.