Case Background
Read dropped O’Keefe, a 16-year veteran of the Boston Police Department, off at the home of his colleague, Brian Albert, in January 2022 after the pair had been drinking at a bar late into the evening. Early the next morning, O’Keefe was discovered dead in Albert’s snow-covered front yard.
Prosecutors accused Read—who was charged in June 2022 with second-degree murder, manslaughter while operating under the influence of alcohol and leaving the scene of personal injury and death—of striking O’Keefe with her car before driving away and leaving him in the snow. Read denied the allegations for years, arguing investigators were attempting to frame her for her boyfriend’s death.
Over the course of two trials, attorneys presented conflicting evidence, including Apple Health data from O’Keefe’s phone. Defense attorneys argued the data showed activity after Read had already left the scene, undermining the prosecution’s theory that she struck him with her vehicle.
No one has been charged with O’Keefe’s murder since Read was acquitted of second-degree murder in June 2025. She was ultimately convicted of driving under the influence of alcohol because her blood alcohol content remained above the legal limit when tested the morning after she returned to the scene where she had dropped O’Keefe off.
Apple Health Data Clashes with Prosecution’s Theory
Apple Health data indicated O’Keefe took dozens of steps after Read dropped him off. Logs showed that seven minutes after he arrived at Albert’s house, the device recorded 36 steps, a distance of just over 80 feet. On the stand, however, a forensic expert testified that such data can have limited accuracy and should be weighed cautiously.
While testing Apple Health data prior to trial, the phone extraction specialist said his own device recorded him climbing nine flights of stairs when he was actually driving uphill with his phone in the car. Apple’s official Health app policy makes clear that the Health app is designed to give users a consolidated view of their personal health information, but it doesn’t make claims about the data meeting legal or medical standards.
A 2019 paper in Digital Investigation found that iPhone step counts closely matched manually measured steps with an average error of about 2% under controlled conditions, suggesting the data can support forensic timelines when paired with stronger evidence. Distances, however, “can deviate up to 30-40% from the true value” due to factors like walking speed and walking style.
Health data was not the only digital evidence introduced at trial. Investigators also extracted location and battery temperature data from O’Keefe’s phone. Prosecutors argued those data points supported their claim that O’Keefe never entered Albert’s home and instead died outside after being struck by Read.
The forensic investigator testified that the phone’s location data placed the device near a flagpole in Albert’s front yard, the same area where O’Keefe’s body was later found. The phone’s battery temperature steadily declined from 77 degrees just after midnight to about 50 degrees by 1:30 a.m. Based on the combined data, the investigator said he believed the phone remained near the flagpole from about 12:30 a.m., when Read dropped O’Keefe off, until 6 a.m., when his body was discovered, supporting the prosecution’s theory that O’Keefe never entered the house.
Digital Forensics Considered Circumstantial
As the case unfolded in court, digital forensics experts weighed how much evidentiary value data like this should carry. Gary Cantrell, an associate teaching professor at Northeastern University with a background in law enforcement digital forensics, said such evidence is typically considered circumstantial.
“It’s very useful and it’s supporting information, but it’s not something you usually build an entire case on because there are many ways to poke holes in the information,” Cantrell told Northeastern Global News in April 2025.
Like Apple Health data, phone location information also has limitations. Cantrell said analysts can often determine whether a device was at a particular residence but may struggle to pinpoint whether it was inside or outside a specific structure.
“I can probably tell that it was at a residence or in a neighborhood,” Cantrell said. “But determining whether a phone was inside the house or outside becomes more challenging the more precise you want to be.”
Cantrell added that photos taken on a device can provide more precise GPS coordinates. Otherwise, phone location data is typically generated through triangulation between cell towers, producing an estimate rather than an exact point.
He also noted a key limitation: There is no way to confirm a phone was physically on a person when it recorded steps or changes in location. For that reason, such data is considered supplemental rather than definitive.
Legal Precedent of Health Data
The use of Apple Health data in the Read case reflects a broader trend in criminal trials, where passive digital records are increasingly treated as objective witnesses.
During the murder trial of Alex Murdaugh, prosecutors leaned heavily on digital evidence from the victims’ phones to dismantle Alex Murdaugh’s alibi. The most consequential evidence was a short video found on his son Paul Murdaugh’s phone, recorded just minutes before he and his mother were killed, which captured Alex’s voice at the family’s dog kennels despite his repeated claims that he was not there. Prosecutors combined this video with step counts and other phone activity data from the victims’ devices to construct a minute-by-minute digital timeline of the crime.
The tension seen in the Read case is likely to shape future prosecutions as courts weigh how much credibility to assign to consumer health data that was never designed for courtroom use. The emerging legal precedent seems to be that health data may be fallible, but it’s admissible, placing the burden on experts to explain its reliability, limitations and potential for misinterpretation.
