Joe Bridges has always been a problem solver. As a mechanical engineer, he spent his career fixing what was broken and improving what could be done better.
That mindset made it even more difficult for Bridges to watch his sister struggle for more than a decade with a medical condition that doctors couldn’t identify. When they finally diagnosed her with acquired angioedema, a rare immune system disorder, he wanted to help other families avoid the same frustration.
“Being an engineer, I didn’t have any choice but to sit down and think about ‘what might I do,’” he said. “Engineers are taught to think logically and solve problems. So, here’s a problem I might be able to solve.”
At 77 years old, he decided to enroll at the UTHealth Houston’s School of Biomedical Informatics to devise a solution. He spent the next three years developing a computer system to help physicians improve the accuracy of their diagnoses.
On Friday, the 80-year-old is graduating as the first in the nation to earn a doctorate in health informatics, a field that focuses on using technology such as artificial intelligence and machine learning to improve the use of data in health care and medical research.
Bridges is just the second person over the age of 60 to earn a degree from the School of Biomedical Informatics since it was founded 23 years ago, said Jiajie Zhang, the school’s dean. The other was a woman in her 60s who earned a master’s degree.
“What he did is just amazing,” Zhang said. “He’s an amazing gentleman and an amazing guy.”
Bridges is among the first group of students who enrolled in the doctorate program, designed as a four-year program for working professionals. Being retired allowed him to take extra classes and graduate early.
His sister, Jan Bridges, has seen a marked improvement since a specialist in San Diego identified the correct diagnosis of acquired angioedema back in 2015. Doctors initially believed she had hereditary angioedema, a similar condition that does not respond to the same treatment. Now she’s no longer experiencing the abdominal pain and swelling that began plaguing her in 2009.
Jan Bridges said she’s humbled by what her brother has done to help others avoid the frustration of a misdiagnosis.
“It’s such a level of deep gratitude, to think that someone could love you that much and be so dedicated to his family and just the general situation,” she said.
Finding a solution
Joe Bridges first became involved at UTHealth Houston in 2011, while his sister was still struggling to find the right diagnosis. He served on the university’s development board and later the advisory councils for the Cizik School of Nursing and School of Biomedical Informatics.
He was serving as chair of the advisory board for the School of Biomedical Informatics when Dr. Giuseppe Colasurdo, the president of UTHealth Houston, suggested he enroll in the new health informatics doctorate program to figure out a way to improve diagnostic accuracy.
While he didn’t have a background in medicine, Bridges saw a similarity to his former career as a mechanical engineer in the oil and gas industry.
“Diagnosis is very similar to the things we do in the oil and gas business,” he said. “I used to tell people ‘you have a patient who has shortness of breath, and I have a gas well that won’t produce enough gas.’”
There are roughly 200 symptoms that can present in patients, but they can be seen in different combinations in approximately 10,000 medical conditions.
Studies have suggested that roughly 5 percent of patients in the U.S. are misdiagnosed at the primary care level. That’s approximately 12 million patients annually.
“It’s not because the doctors are incapable of making a diagnosis,” Zhang said. “If something is rare and they’ve never seen it, not in their medical education and not in their practice, they may need some help.”
Bridges decided the best solution involved the use of a computerized diagnostic decision support system, which is used to help guide physicians toward a diagnosis. He evaluated products already on the market and settled on one that was fast, accurate and able to outperform physicians.
Physicians can use Bridges’ system by inputting the patient’s symptoms and some basic demographic information, such as their age range and travel history. The system combs through data to offer a list of medical conditions that fit those symptoms. The physician then determines the best match.
“What this does is try to put everybody on an even playing field, looking for the conditions that display these clinical features,” he said.
Hope for better diagnoses
Bridges first tested the system through a case study in which 120 residents from the McGovern Medical School at UTHealth Houston evaluated 24 cases. He found that when residents used the system, there was a 9.1 percent improvement in diagnostic accuracy.
He then asked UTHealth physicians to test the system in a live clinical setting. The data he collected showed that 78 percent of physicians thought the system was useful, while 36.8 percent said the system prompted them to change their initial diagnosis to something that was a better match.
Now that Bridges has developed a system that shows promise, he’s hopeful it can be put into practice. He plans to present it to other UTHealth institutions in the hopes that more physicians will be persuaded to use it. He’s ultimately hoping to find a “clinical champion” in a respected doctor who will be able to generate more interest through word of mouth.
Jan Bridges is optimistic that her brother’s system will help others avoid the frustration she experienced.
“I’m hoping this will move diagnosis for patients who have what I have, and other diseases, to a much shorter journey than we took,” she said.
Regardless, Bridges is satisfied with the work he’s completed so far. It was difficult for him to return to academia at 77 years old, but he’s grateful he was able to study and learn, particularly because it helped him address an issue that was so personal to him.
“I don’t suggest to anybody that what I’ve done will solve the problem of diagnostic inaccuracy. I think that’s way too much to hope for,” he said. “But I do think that this can introduce a concept and mechanism for improving accuracy, reducing diagnostic inaccuracy, and that will help a lot of people over the years if we can get it embedded in practices.”
link