Jocelyn Escanuela is on Walmart’s checkout line when an unknown number pops up on the phone. She still couldn't explain why she picked it up and then listened to a puck tone that sounded like a scam.

The caller told her that she was selected to receive a $6,000 grant. She will have a personal assistant to help her through the "crisis".

How did they even know she was in crisis?

It turns out that the caller is legal. She comes from the Homeless Prevention Unit, an experimental Los Angeles County program, which is testing whether it can start by taking them out of the hills to stop homelessness before people (one person) begin at a time.

Escanuela's crisis is not discovered by one person, but a predictive statistical model that addresses the challenge of preventive measures that make homeless is an attractive but unused strategy.

While there is evidence that services such as deportation of defense and financial aid can prevent people from becoming homeless, in fact, it is impossible to know if any given person will become homeless without help. Research shows that only a small percentage. The elusive goal of prevention is to determine that this proportion is small.

“With limited prevention resources, there are real consequences with the people they need them the most,” said Steve Berg, chief policy officer of the National Alliance, ending homelessness, which has historically been frowning on expensive prevention programs.

But Berger said: "If these emerging technologies prove to be effective in predicting who is most likely to become homeless, that would be good news."

As both the city’s ULA “luxury home tax” and county-wide measurement business tax both begin to direct millions of dollars in prevention to homeless prevention, it will become increasingly important to achieve elusive precision.

Escanuela is being tested as a high-risk model to understand its effectiveness.

It was created by the California Policy Laboratory of California at the University of California, Los Angeles (UCLA), a research institute that studies data from all visiting county agencies, such as the Department of Health and Social Services, that interact with their most vulnerable people. The Policy Lab screened through all data and evaluated approximately 500 markers to generate a list of individuals and families predicted by its model that predicted the high risk of homelessness. It shifted the list to the homeless preventive unit and its housing stability team.

“When people just leave the hospital, we meet them, when they just lose their jobs, we meet them. “When people lose their only family members of their suppliers, we meet them.” We meet people who receive verbal eviction warnings from their landlords. ”

Homeless Prevention Unit Analysts randomly worked in the name on the high-risk list to propose two groups of candidates. Half of the intervention will be provided - cash allowance and case manager for four months. The other half will get nothing and will never know they are selected, but will be monitored by any contact with the county or homeless agency they have established.

Escanuela landed in the target population of randomized clinical trials – the lucky half.

The Holy Grail of Prevention will be a model that identifies those who will be homeless and avoid spending money on those who will never.

In a 2023 report, the University of Notre Dame’s Economic Opportunity Laboratory found that people served in Santa Clara County’s prevention program were nearly 80% more likely to be homeless than the control group after receiving services.

This doesn't sound impressive, as only 4.1% of people who don't help are homeless, suggesting that investing a lot of money in people who don't become homeless without help is.

"It is possible to predict even if not very good," said Beth Shinn, a research professor at Vanderbilt University.

Her research found that the model performed better than predicting outreach workers.

"Even the way the city develops, it is cost-effective and moderately successful," Shinn said.

These studies involve people seeking preventive services. Policy labs and homeless prevention departments are taking the next step to use predictive analytics to find people who have not yet sought services.

Early discovery is promising. Of the data used to build the model, about 47,000 people were served by county, and 24% of those expected to be at high risk were actually homeless, compared with only 7% of the entire sample.

It turns out to be effective in finding people who may be homeless for a long time.

"Our customers have very high risks," Vanderford said. "They have complex health and mental health conditions. They meet with us in real moments of crisis. The timing of our contact with customers seems magical to me."

After completing the four-month plan, enough people have been tracked for 18 months until 2027, the full results of the trial will not reach the final result.

Homeless Prevention Units are funded by the U.S. Rescue Program Act and are supplemented by county funds. It has about 250 active customers, and its turnover is four to six months and can cope with 750 per year. Vanderford said about 90% of homes have been retained or new homes have been found.

This is labor-intensive work. Four analysts browsed the original list of randomly filtered unqualified candidates. Many of the people on the final list are already homeless, demonstrating the accurate predictions due to the delays in obtaining county data from the Policy Lab.

"There is a real challenge to stay in touch with people," Van derford said. "The phone goes out. The client may be hospitalized or jailed. The client may not trust the phone, which sounds a little too good to achieve. There is no doubt about voicemail."

“I’ve never answered this way,” Escanuela said. "I don't know what forced me to answer."

Neither Escanuela nor Vanderford knew what specific factors put her on the high-risk list, except that she was visiting county services.

But the phone was timely. She and her family are in a protracted battle of expulsion, and she fears being homeless again.

She said she spent a long time in the park and at night in church as a child, and later lived in a joint rescue mission in downtown and in the Hope Gardens Shelter in Silma.

"I don't want to go back to this, especially adults," she said.

After enrolling, Escanuela received a call from Chris Schucchert, one of the program's 20 contract case managers. During the four-month and two-month extensions of the program, they communicated by phone and text. He helps in a variety of ways, from grocery debit cards to meeting her emotional needs.

"Chris was able to find my therapist," Escanuela said. “I went through a lot when I just needed someone to talk.”

The program’s case manager handles client fees and pays directly to suppliers and landlords. The case manager must apply, which is usually after negotiations with the client. Schuchert says that state items like $250 shoes boil down to $50 models, but a $800 bed can be approved for a better sleeping bed.

Case Manager Recommends to Health and Mental Health Agency. When clients owe thousands of dollars on rent, they can refer them to Los Angeles where the accommodation is and other groups that help encounter troubled tenants.

In the case of Escanuela, this is not necessary because the landlord stops accepting rent during the eviction process. Schuchert believes it would be better to let her leave the apartment she shared with her mother and brother to avoid eviction records.

Today, she lives with her mother in a pleasant apartment in Pomona. She saved enough money to pay for moving and bought equipment for home businesses that provide eyelash services.

She said the business was good enough that she was paying the money on her own.