Image by 123RF
By Shiraz Austin, MD, and Co-Founder of Augnito, a cloud-based voice AI technology, discusses why speech recognition is vital for radiology today
The scope of radiological imagery has come a long way since 1895 when the first x-ray scanning processes were used for patient diagnostics. Today, radiologists have access to a continuously growing number of imaging techniques and methods that have been introduced over time, including magnetic resonance imaging, molecular techniques, and ultrasound, amongst others.
Yet where the clinical radiologist’s primary function is to diagnose and monitor the progression of some of the world’s most complex diseases and injuries, their findings need to be documented accurately and precisely to ensure their patients access the most efficient care. Of course, healthcare professionals in this field create highly complex reports, often with multiple images and comprehensive studies of each one making up just one report. It is no wonder that radiology is one of the most highly skilled and highly paid healthcare institutions. Yet, at the same time, they are often open to litigation if mistakes or errors are detected, meaning absolute accuracy, precision, and reliability are of critical importance in radiology reporting.
The challenge of finding time for patient care
At the heart of every healthcare professional’s working day is their need to deliver the highest level of care for each patient. The healthcare professional must record relevant and timely information and detailed and accurate descriptions that make up the patient report. A more precise diagnosis and documentation will enable doctors to spend more time with the patient, doing what they do best.
Yet the challenge widely held by all radiologists is finding the right amount of time available to carrying out and finishing tasks, such as completing clinical reports or documents and waiting for transcription, which takes a lot out of their already pressured schedule. Over time, this challenge intensifies as workflows become more complex, which adds to the ever-increasing pressure radiologists face across the country today.
So, how can these needs be met to ensure the priority of direct patient care is not compromised?
Speech recognition software for effective healthcare
Modern speech recognition (SR) software allows radiologists to create documents in real-time and faster than traditional manual transcription. The software can record dictation and stream the audio as the speaker dictates the words. This technology has already proven to put an hour back into a healthcare professional’s working day, every day, limiting the risk of errors and misinterpretations. It has also been proven that SR software can directly impact reducing lengthy patient admissions and post-treatment, thus seeing the patient’s return to the community become a smoother and faster process due to the reduced turnaround time.
However, it is essential to note that this technology has advanced over many years. Traditional SR relied on a more straightforward process, being installed on premise / on device, resulting in a slower and less accurate transcription level, causing the radiologists to re-read, check, and double-check before signing off the completed report. For example, when frequent errors were made by traditional on-premise / on device SR technology, the author consistently tried to correct an individual word or phrase. This would often lead to user profile corruption due to previous pronunciation attempts, meaning that the author would need to re-create the profile again.
In addition, the technology historically also relied on the author speaking coherently in a quiet environment, often using an out-of-date workstation PC located far away from the hospital’s limited IT resources. When a healthcare professional is up against a hardware system that is both outdated and limited in its use with newer software, it can often be an obstacle for advancement.
These challenges have now been overcome with the arrival of AI-driven, cloud-based SR solutions. Of critical importance to radiologists, recent advancements to this technology have significantly improved clinical notetaking and the development of patient records, including accuracy in voice dictation and the impact of accents or dialects on word recognition, which also reduce the clinical risk in errors and misinterpretations previously encountered.
Using AI-driven voice transcription, the healthcare professional can drive software, applications, or devices that provide quicker outputs with fewer compromises. Gone are the days of outdated workstations and limited hospital IT resources which have largely dictated workflow. Thankfully, today’s technologies are becoming increasingly visible, and constantly evolving, with voice dictation set to be our standard primary method of interaction with computers.
The benefits of AI-driven speech recognition in radiology today
Radiologists are in short supply, with the NHS cited to be short of almost 2,000 professionals in this field. According to census stats by the Royal College of Radiologists, it is feared that the figure could hit around 6,000 by 2030. As such, the involvement of speech recognition in this field has a crucial role.
Radiologists spend around a third of their working week reporting, and juggling departmental requirements with this likely to increase in the future, as medicine naturally advances and the aging population lives longer with a better chance of successfully living with chronic diseases or surviving illnesses such as cancer. Examinations will grow more complex, as will the ongoing issue of limited resources and hiring new staff.
To manage their time more effectively, radiologists need dependable technology that’s out of the box ready to support medical vocabularies. They rely on robust, quantitative documentation, which needs to be fast and accurate to meet deadlines. More importantly, documentation needs to adjust to the way a radiologist works. For example, it needs to integrate seamlessly into an existing workflow, even in the busiest and noisiest of working environments. This is where AI-driven, cloud-based SR solutions can be a real game-changer.
Software available to healthcare professionals has been purchased on a tight budget, but the future is a growing concern. With the increasing costs in image technology and a slow uptake into the profession by the current generation of radiologists, SR technology needs to be prepared to ride the wave of demand, meaning speech software engines must perform as well as the market expects. To do this, it needs to be continuously improving its accuracy. Such technology can only contribute to a greater chance of a work-life balance for those coming into the profession and, in turn, offer a more efficient and productive care to those who need it most.