The Homeless Link charity asks members of the public to flag the locations of rough sleepers using its StreetLink app so that it can alert the relevant local authority or outreach group to send help.
But the time it takes to manually sort through the results means that many have moved on by the time help can be found, and currently only about 15% are located and offered support.
Now a team of students and tech workers on a data science fellowship have developed a tool that will flag the most useful reports and increase the number of people offered help.
"It could be very helpful," Matt Harrison, the director of StreetLink told the Thomson Reuters Foundation.
"If we can prioritise our scarce resources on the alerts which are mostly likely to result in a homeless person being found and helped then we should do that," he said.
Thousands of people sleep rough on England's streets every night, and they have become an increasingly common sight in major cities including London, Birmingham, and Manchester.
But with up to 1,000 reports arriving each day in peak times such as winter, it can take several days to manually sort the reports and pass them on, by which point in many cases the homeless person will have moved on.
At present, just 14% of homeless people reported sleeping rough through StreetLink are ultimately found by outreach workers, with the charity saying delays, duplicated reports, and incomplete information all contributing to the low success rate.
A 4-strong team taking part in the Data Science for Social Good scheme worked with Homeless Link to help streamline their response by developing an data tool that identifies the most useful reports.
"Reaching folks as soon as possible is really critical, especially when conditions are really harsh outside," team member Austin Nguyen told the Thomson Reuters Foundation. "The earlier you can support someone before they get into a chronic patterns of sleeping on the streets, the better the outcomes are."
