Digital Data and the Spoken Word: Bridging the Communication Gap
As radiology equipment becomes ever more advanced, so-called “structured data” only increases. Advanced 3D and reconstruction visualization generates measurements from multiple angles, and the digital information builds and builds. 
Seamlessly combining structured data with the spoken word can be tricky. Mistakes can cost time, money, and worst of all can lead to poor patient care. Juergen Fritsch, PhD, chief scientist and co-founder of Franklin, Tenn-based M*Modal, devotes most of his waking moments figuring out the best ways to integrate voice and data.
These days, it’s relatively easy to capture structured data, import it, and store it. The art is using technology, combined with human workflows, to avoid mistakes and boost efficiency. “We link structured data with the spoken documentation,” explains Fritsch, who operates from an office in Pittsburgh. “A radiologist may say, ‘This measurement has grown since the last time I saw this patient.’ With our technology, specifically with natural language understanding technology, we parse the verbiage that they dictated—and either identify new structured data measurements—or link them up with existing ones.”
If the spoken word conflicts with existing digital measurements, the M*Modal system tells the radiologist. “If they are talking about a lesion in the left lung, for example, we pair up that dictation with the actual measurements,” Fritsch adds. “There may be a contradiction. The radiologist might say, ‘This lesion has diminished,’ and we notice from actual measurements that it did not diminish. In fact, it actually grew. Or the radiologist accidentally refers to the right instead of the left lung. We can bring it up as an alert and inform the radiologist that the data contradicts the dictation. That is the real challenging aspect that makes the biggest difference. We put the data in proper context.”
Minimizing Errors
Dictating measurements into a microphone always carries the potential for error. The M*Modal solution is to automatically integrate the electronically measured data into the radiologist’s report.
These days, measurements go beyond lesions to encompass various clinical aspects that must be properly documented. “In CT imaging, patients sometimes get a contrast agent injected into the body,” says Fritsch. “The equipment measures the flow rate, and the amount that goes into patients is captured, but it’s not typically integrated with other systems. Again, there is the risk that the data gets lost or is not properly relayed, interpreted, or dictated. That discrepancy is a risk for patients, but we are able to eliminate that risk.”
In most parts of the country, the radiology report is still dictated. This process of creating a narrative is still a vital side of the profession, and preserving that human element—while avoiding mistakes—is the goal at M*Modal.
Government entities, on the other hand, are increasingly enamored with data collected from devices, or directly from physicians. This push toward structured data is largely driven by meaningful use rules, which essentially award money to practices that successfully adopt electronic systems.
Unfortunately, hastily adopted systems can be tedious to use, leading to slowdowns and a lack of integration. “We help by making it easier and more efficient for physicians to capture the necessary documentation for meaningful use,” enthuses Fritsch. “Rather than making users point and click or use drop down reporting systems, which typically take two to three times longer, we allow them to dictate—just narrate. Our system identifies the data in the narrative, supporting practices in fulfilling government requirements and complying with meaningful use.”
Maintaining patient care while increasing speed and profitability is not easy, but Fritsch believes it’s possible, even necessary, in today’s environment. “It’s all about efficiency in the radiology space,” he says. “If you can do it in a minute, rather than a minute and 10 seconds, it adds up to a huge savings, possibly allowing you to see more patients in a day.”
Other clinical documentation systems, he contends, may not document all the required elements. When that happens, radiologists may incorrectly believe they are in compliance. “So we also provide feedback to radiologists and users that either something is missing in the report, or meaningful use guidelines require them to be more specific about something,” says Fritsch. “We analyze the dictation and give them feedback that allows them to be compliant. It’s more than just faster and easier, it is complete and compliant.”
When everything works as it should, the complete package amounts to a melding of the human element and the best in technical ingenuity. “We rely on this combination of the technology taking a first crack at it, and the human labor force potentially reviewing it and correcting it,” explains Fritsch. “That is primarily happening in the non-radiology specialties. Radiologists are a little different, because radiologists sit in a quiet environment in a reading room. Their recordings are typically high quality with very little noise. For the most part, there is little or no labor involved in correcting or proofreading their dictations. They do it themselves, so it’s quicker and more real time.”
Humble Beginnings
Little more than a decade ago, Fritsch co-founded M*Modal with a great idea and a driving ambition to transform clinical transcription services. Today, the 12,000-employee international organization has expanded into clinical documentation workflow solutions, advanced cloud-based Speech Understanding™ technology, and advanced unstructured data analytics.
Fritsch and two fellow graduate students at Pittsburgh-based Carnegie Mellon University first gained traction by winning a national technology competition funded by the United States government. The award winning software was designed as a cloak-and-dagger method to listen in on certain conversations.
“We noticed that radiologists and other physicians were not that different from people talking on the phone when they are documenting clinical encounters,” muses Fritsch. “They are not necessarily grammatical. They occasionally mumble. We figured the same technology would apply, and it worked beautifully.”
Fritsch and his colleagues ultimately started M*Modal using new technology developed from the ground up to avoid any ties to former projects. From the beginning, the founders knew that excellent software was not enough if it did not address the particular problems of users. “Success depends on having people in the field who understand the specific needs of radiologists,” says Fritsch. “These analysts and product managers collaborate with customers to understand what they need and how they need it.”
Of the 12,000 people in the M*Modal work force, about 10,000 are working to transcribe, edit, and correct data sent to them. A huge workflow engine takes physician documentation in the form of audio recordings and turns them into structured documents. The mix of technology and labor is essential because Fritsch points out that “the technology is never perfect and will never produce a 100% accurate report—at least not in the typical environment.”
Always trying to improve the “user experience” has helped M*Modal succeed in a competitive environment. Despite high customer satisfaction ratings, Fritsch expresses a fundamental wish to improve on all fronts. “We spend a lot of resources on the user experience,” he says. “We want to simplify the systems. With an Apple product, you can just open it and figure out how to use it because it is fairly intuitive, and that’s where we need to get with these health care systems. We are not there yet. We must simplify our systems and make them more usable and more intuitive.”
Fritsch watches his competitors closely, but the task becomes easier by talking to people with experience using the competitor’s product. The fact that tough competitors exist is proof of a vibrant market, and it’s a market condition that spurs M*Modal to keep innovating.
Ultimately, Fritsch places most of his considerable powers of concentration on cracking the efficiency nut in ever more creative ways. “The idea is to fulfill all these various requirements, because that’s how users can improve their return on investment,” he concludes. “There is a shortage of radiologists, and will be for quite some time. That speaks to the need for efficiency. Radiologists must be able to cover the volumes with the people they have, and M*Modal can help them do that.” Greg Thompson is a contributing writer for imagingBiz.
Seamlessly combining structured data with the spoken word can be tricky. Mistakes can cost time, money, and worst of all can lead to poor patient care. Juergen Fritsch, PhD, chief scientist and co-founder of Franklin, Tenn-based M*Modal, devotes most of his waking moments figuring out the best ways to integrate voice and data.
These days, it’s relatively easy to capture structured data, import it, and store it. The art is using technology, combined with human workflows, to avoid mistakes and boost efficiency. “We link structured data with the spoken documentation,” explains Fritsch, who operates from an office in Pittsburgh. “A radiologist may say, ‘This measurement has grown since the last time I saw this patient.’ With our technology, specifically with natural language understanding technology, we parse the verbiage that they dictated—and either identify new structured data measurements—or link them up with existing ones.”
If the spoken word conflicts with existing digital measurements, the M*Modal system tells the radiologist. “If they are talking about a lesion in the left lung, for example, we pair up that dictation with the actual measurements,” Fritsch adds. “There may be a contradiction. The radiologist might say, ‘This lesion has diminished,’ and we notice from actual measurements that it did not diminish. In fact, it actually grew. Or the radiologist accidentally refers to the right instead of the left lung. We can bring it up as an alert and inform the radiologist that the data contradicts the dictation. That is the real challenging aspect that makes the biggest difference. We put the data in proper context.”
Minimizing Errors
Dictating measurements into a microphone always carries the potential for error. The M*Modal solution is to automatically integrate the electronically measured data into the radiologist’s report.
These days, measurements go beyond lesions to encompass various clinical aspects that must be properly documented. “In CT imaging, patients sometimes get a contrast agent injected into the body,” says Fritsch. “The equipment measures the flow rate, and the amount that goes into patients is captured, but it’s not typically integrated with other systems. Again, there is the risk that the data gets lost or is not properly relayed, interpreted, or dictated. That discrepancy is a risk for patients, but we are able to eliminate that risk.”
In most parts of the country, the radiology report is still dictated. This process of creating a narrative is still a vital side of the profession, and preserving that human element—while avoiding mistakes—is the goal at M*Modal.
Government entities, on the other hand, are increasingly enamored with data collected from devices, or directly from physicians. This push toward structured data is largely driven by meaningful use rules, which essentially award money to practices that successfully adopt electronic systems.
Unfortunately, hastily adopted systems can be tedious to use, leading to slowdowns and a lack of integration. “We help by making it easier and more efficient for physicians to capture the necessary documentation for meaningful use,” enthuses Fritsch. “Rather than making users point and click or use drop down reporting systems, which typically take two to three times longer, we allow them to dictate—just narrate. Our system identifies the data in the narrative, supporting practices in fulfilling government requirements and complying with meaningful use.”
Maintaining patient care while increasing speed and profitability is not easy, but Fritsch believes it’s possible, even necessary, in today’s environment. “It’s all about efficiency in the radiology space,” he says. “If you can do it in a minute, rather than a minute and 10 seconds, it adds up to a huge savings, possibly allowing you to see more patients in a day.”
Other clinical documentation systems, he contends, may not document all the required elements. When that happens, radiologists may incorrectly believe they are in compliance. “So we also provide feedback to radiologists and users that either something is missing in the report, or meaningful use guidelines require them to be more specific about something,” says Fritsch. “We analyze the dictation and give them feedback that allows them to be compliant. It’s more than just faster and easier, it is complete and compliant.”
When everything works as it should, the complete package amounts to a melding of the human element and the best in technical ingenuity. “We rely on this combination of the technology taking a first crack at it, and the human labor force potentially reviewing it and correcting it,” explains Fritsch. “That is primarily happening in the non-radiology specialties. Radiologists are a little different, because radiologists sit in a quiet environment in a reading room. Their recordings are typically high quality with very little noise. For the most part, there is little or no labor involved in correcting or proofreading their dictations. They do it themselves, so it’s quicker and more real time.”
Humble Beginnings
Little more than a decade ago, Fritsch co-founded M*Modal with a great idea and a driving ambition to transform clinical transcription services. Today, the 12,000-employee international organization has expanded into clinical documentation workflow solutions, advanced cloud-based Speech Understanding™ technology, and advanced unstructured data analytics.
Fritsch and two fellow graduate students at Pittsburgh-based Carnegie Mellon University first gained traction by winning a national technology competition funded by the United States government. The award winning software was designed as a cloak-and-dagger method to listen in on certain conversations.
“We noticed that radiologists and other physicians were not that different from people talking on the phone when they are documenting clinical encounters,” muses Fritsch. “They are not necessarily grammatical. They occasionally mumble. We figured the same technology would apply, and it worked beautifully.”
Fritsch and his colleagues ultimately started M*Modal using new technology developed from the ground up to avoid any ties to former projects. From the beginning, the founders knew that excellent software was not enough if it did not address the particular problems of users. “Success depends on having people in the field who understand the specific needs of radiologists,” says Fritsch. “These analysts and product managers collaborate with customers to understand what they need and how they need it.”
Of the 12,000 people in the M*Modal work force, about 10,000 are working to transcribe, edit, and correct data sent to them. A huge workflow engine takes physician documentation in the form of audio recordings and turns them into structured documents. The mix of technology and labor is essential because Fritsch points out that “the technology is never perfect and will never produce a 100% accurate report—at least not in the typical environment.”
Always trying to improve the “user experience” has helped M*Modal succeed in a competitive environment. Despite high customer satisfaction ratings, Fritsch expresses a fundamental wish to improve on all fronts. “We spend a lot of resources on the user experience,” he says. “We want to simplify the systems. With an Apple product, you can just open it and figure out how to use it because it is fairly intuitive, and that’s where we need to get with these health care systems. We are not there yet. We must simplify our systems and make them more usable and more intuitive.”
Fritsch watches his competitors closely, but the task becomes easier by talking to people with experience using the competitor’s product. The fact that tough competitors exist is proof of a vibrant market, and it’s a market condition that spurs M*Modal to keep innovating.
Ultimately, Fritsch places most of his considerable powers of concentration on cracking the efficiency nut in ever more creative ways. “The idea is to fulfill all these various requirements, because that’s how users can improve their return on investment,” he concludes. “There is a shortage of radiologists, and will be for quite some time. That speaks to the need for efficiency. Radiologists must be able to cover the volumes with the people they have, and M*Modal can help them do that.” Greg Thompson is a contributing writer for imagingBiz.