Robertson Research Institute

This Physician’s Assistant Never Sleeps

The emergency room doors slammed open and the paramedics rushed in. Their patient—a young man who, at 21, should have had most of his life ahead of him—was suffering from violent seizures. As if that weren’t enough, he was also in a coma.

The emergency room team raced to diagnose his condition and stabilize him. Early tests provided an answer—and more questions: The young man had kidney failure. That would explain the symptoms, but what would explain the kidney failure and how to treat it? The doctors labored for three days to find those answers. And then their patient died.

In his grief, the young man’s father turned to a longtime friend, Michigan neuropharmacologist Dr. Joel C. Robertson, to try to make sense of the tragedy. Robertson reviewed the medical records and researched the medical literature. The young man’s doctors had been focused on the most likely explanations for his symptoms—but that focus blinded them to an inconsistency in the test results that should have ruled out those diagnoses. Had they reflected on all possible causes, they would have considered a rare metabolic condition that was consistent with the test results. As it turned out, the young man had that rare condition, which had remained undiagnosed until after his death.

Robertson presented his findings to his friend and to the doctors, who acknowledged that if they’d had Robertson’s research in the emergency room, the young man might have lived. But Robertson had taken two weeks to ferret out the correct diagnosis; neither the doctors nor their patient had had the luxury of that much time.

The tragedy gave Robertson his first experience with the fatal consequences of misdiagnosis, but the incident, sadly, is far from unique. Nearly 100,000 people die in U.S. hospitals each year due to diagnostic errors, according to the Institute of Medicine, making misdiagnosis one of the top 10 causes of death in the United States—and, amazingly, a more common cause of death than car accidents, breast cancer, or AIDS, according to the Agency for Healthcare Research and Quality.

The problem isn’t limited to the United States and other industrialized countries. In developing countries, doctors often must make diagnoses without the benefit of the labs, equipment, and sophisticated tests—readily available to doctors in much of the developed countries—that promote accurate diagnoses. In the rural areas of many countries, the scarcity of doctors means that diagnoses are often made by healthcare professionals without medical degrees and training. Both factors contribute to diagnostic errors and fatalities worldwide.

Robertson’s experience touched him deeply and quickly led to a personal mission: to turn the tragedy into an opportunity to help others by reducing the global rate of death from misdiagnoses. Robertson brought the idea to his friend, who agreed to help fund it. The Robertson Research Institute, which Robertson founded and headed, would turn the idea into reality.

“My vision was of a truly global medical solution—one that would save lives worldwide through faster and more accurate diagnoses, regardless of the level of medical resources available to the healthcare providers and regardless of the language or culture of those providers and their patients,” says Robertson. “There were medical diagnostic programs already on the market—but none did what I envisioned. They were not designed to be easy for physicians to use under emergency situations. They were not designed for use in developing countries, where expensive computers and Internet access might be unavailable. They were little more than digitized versions of medical texts, inflexible and outdated by the time they were released.

“I envisioned a solution that would work on whatever platform was available to the healthcare provider—a desktop PC in teaching hospitals, maybe a Tablet PC or Pocket PC in remote locations,” says Robertson. “For maximum flexibility in unfavorable circumstances, it would have to work with small memory requirements and without Internet connections. Yet it had to support vast amounts of text, photos, illustrations, and other data. It had to be highly extensible, so that we could continually update the database with thousands of symptoms and diagnoses; expand to support more languages and cultures—such as Eastern medicine, which is very different from what we practice in the West; and, ultimately, evolve from an emergencycare diagnostic tool to a solution to support wellness, disease diagnosis, and virtually all other aspects of medical knowledge.

“The question I had was this: Does the technology exist that can turn this vision into a reality?” recalls Robertson.

Rob Cecil, at the time a senior architect at Robertson Research Institute, worked with Microsoft® Gold Certified Partner Sagestone Consulting, now part of NuSoft Solutions, to answer that question. They rejected a popular technology choice—the Java development system—because they estimated they’d need to write too much software code for the product to be efficient and because they wouldn’t be able to get the software to run easily on the various types of computers that Robertson wanted healthcare providers to be able to use.

We’re ready to save lives now.  Dr. Joel Robertson CEO Robertson Technologies

Cecil and his colleagues thought one technology held promise: Microsoft Windows® and .NET software for connecting people, information, systems, and devices. With this solution, Cecil and his colleagues decided they could create their product once and have it run on various types of computers. They could create software that could run independently on Pocket PCs and Tablet PCs when those machines were far from a computer network or the Internet—say, in an isolated, rural area in a developing country—yet also receive new medical information from computers at the Robertson Research Institute when Internet connections were available. Cecil and the development team could use Microsoft technologies as building blocks to create their software quickly, and then rearrange or add components to change or expand the software over time.

The second hurdle that Cecil and his colleagues had to overcome was how to create software that would be more than a digitized version of medical texts, but in fact would lead healthcare professionals to correct diagnoses that they might not otherwise have considered. To achieve this, the software would have to be able to analyze symptoms, weigh the possible diagnoses, and know the best questions to ask to confirm or refute those diagnoses.

The Robertson team built software to deliver diagnoses—and then watched in horror as it didn’t work. For a fledgling organization with limited capital, the setback was significant. “We used an approach that everyone recommended, and it nearly sank us,” recalls Robertson. “We were running out of money and out of options.”

Just when they needed it, the Robertson developers got help from an unlikely source: a Presbyterian minister named Thomas Bayes— who’d been dead for some 240 years. Back in the eighteenth century, Bayes, who was also an amateur mathematician, had tackled the problem of evaluating the likelihood of various outcomes based on prior observations, exactly the problem now confronting Robertson, Cecil, and their colleagues. Bayes may have been long gone, but Dr. David Heckerman, one of the foremost experts in Bayesian theory and a leading researcher at Microsoft Research, was very much alive. He guided the Robertson developers in creating software, again using Microsoft technologies, that applied Bayesian theory to the problem of producing accurate diagnoses.

The result of all this effort is NxOpinion, a product that fulfills Robertson’s original vision of delivering faster and more accurate medical diagnoses, even in remote locations with limited medical resources. Trials of NxOpinion are now underway in the Dominican Republic and the Democratic Republic of the Congo, and are scheduled soon for India. Based on the positive test results and feedback he’s received so far, Robertson expects NxOpinion to be released for general use in developing countries and isolated environments by the middle of 2007.

“The product helps me to give my patients the best care possible,” says Dr. Edwin Disla, who’s completing an internship at the Centro Cristiano de Servicios Médicos hospital in the rural outskirts of Santo Domingo, the capital of the Dominican Republic. Disla has participated in the NxOpinion trial there since December 2005.

“When you have little more than a stethoscope in a rural clinic, you find yourself fishing for answers,” says Disla. “I use NxOpinion with most of my patients, generally when I’m presented with several types of symptoms that don’t seem to make any sense. It gives me potential diagnoses I wouldn’t have thought of on my own—and then it suggests tests to confirm the correct diagnosis and even suggests treatment. The doctors I know outside of this hospital all want a copy of the software.”

NxOpinion prompts a doctor to enter details about a patient’s condition as if the doctor were describing the case to a colleague. The software accepts either standard medical language or lay terms to accommodate varying levels of medical knowledge. Weighing each new piece of information, NxOpinion suggests possible ailments and the best question to ask next to narrow the possibilities further. Sometimes those questions seek additional observations or symptoms.

Sometimes the software suggests tests that can help lead to a diagnosis. And when an expensive test, such as an MRI or a CAT scan, isn’t feasible, NxOpinion can suggest lowercost alternatives, such as an enzyme test costing about U.S.$150 that may be the nextbest option.

“I think NxOpinion is the best idea to come along in a long time,” says Dr. Paul Groen, founder of Doctors On Call for Service (DOCS), a nonprofit foundation that sends volunteer U.S. doctors worldwide to help promote global medical care. “I have worked many years in Africa in places where there are no resources—books or journals—and no means of communication. This project is a godsend.”

There’s a 35-year-old man outside of Santo Domingo who would agree. The man, call him “Carlos,” was racked with pain when he came to see Disla in the emergency room at the hospital. He had pain down the right side of his neck and in his head, plus eye irritation and muscle weakness. His symptoms had continued unabated for six months, affecting his ability to work at a local meat factory and, consequently, reducing his already modest income.

The product helps me to give my patients the best care possible.  Dr. Edwin Disla Intern Centro Cristiano de Servicios Médicos

Carlos had seen other doctors but none had been able to offer a diagnosis. His meeting with Disla proved no different. The doctor brought in three of his colleagues for a consultation, but the best they could do was send Carlos home with an antiinflammatory medication, a recommendation to rest, and a slip to return to the hospital for another appointment the following week.

Carlos couldn’t have known it at the time, but his luck—and his health—was about to change. That afternoon, the first PCs running NxOpinion were brought into Disla’s hospital, and Disla was one of the doctors trained on the new software. Intrigued by its possibilities, Disla immediately entered Carlos’s information to see what NxOpinion would suggest.

He was startled when NxOpinion suggested thoracic outlet syndrome, a neurological disorder. The disease is seldom seen in the Dominican Republic, but as Disla read NxOpinion’s description of the disorder, he realized that it caused the type of pain and nerve damage that Carlos exhibited—and it was caused by repetitive activities, the type of activities that a meatfactory worker like Carlos would likely perform. In fact, Disla found out that Carlos constantly used his right arm to pull meat onto the cutting table.

NxOpinion was suggesting a type of obscure orthopedic condition with which Disla could not have been expected to be familiar Thanks to the diagnosis of NxOpinion, Disla referred Carlos to an orthopedist, who treated him successfully. Carlos got back his livelihood and his life.

As work on NxOpinion has continued, the possible uses for the software have grown beyond Robertson’s original vision. The flexibility of the .NET technology at its core means that Robertson can easily modify NxOpinion and use its proprietary Bayesian software for other purposes. To capitalize on these possibilities, Robertson created a forprofit company called Robertson Technologies, of which he is Chief Executive Officer. Its President, Sally Oyster, describes the variations currently under development, including tools for medical training, managedcare cost containment, patient records, pharmaceutical testing—even as a tool to help minimize medical malpractice liability.

“Early on, we saw the possibility of helping farflung healthcare workers who might not have full medical training,” says Oyster. “But we’ve come to realize that NxOpinion can also provide backup to medical specialists operating outside of their specialties, or even to any physician who can use a backup diagnostic tool to help limit malpractice liability. We’re adding a tracking mechanism so peerreview committees can easily see whether correct steps were taken in patient care.”

“We’ll partner with other companies to introduce some of these other versions of NxOpinion over the next year or so,” says Robertson, who now has the alltoorare pleasure of seeing a dream become a reality. “But first we’ll get it out to the people in the developing world who need it. We’re ready to save lives now.”

For more information about Microsoft .NET, visit:
www.microsoft.com/net/default.mspx  

For more information about NxOpinion, see the Microsoft case study at:
<a href="http://www.microsoft.com/casestudies/casestudy.aspx?casestudyid>http://www.microsoft.com/casestudies/casestudy.aspx?casestudyid=52508    

© 2006 Microsoft Corporation. All rights reserved.

 

This document is for informational purposes only. MICROSOFT MAKES NO WARRANTIES, EXPRESS OR IMPLIED, IN THIS SUMMARY.