The Potential of Artificial Intelligence in CI Programming : The Hearing Journal


Artificial intelligence (AI) has found ubiquitous applications in health care, from prediction models that warn clinicians of high-risk conditions like sepsis and heart failure to programs offering tailored cancer treatment recommendations based on patients’ genetic profiles.2 Now, that list includes an AI cochlear implant (CI) programming software tool that shows promise as a faster and more standardized approach to the fitting of the device.

A group of researchers, Govaerts, et al.,2 in Antwerp, Belgium, introduced the concept of target-driven programming, which is accomplished by setting specific psychoacoustic targets, or defined outcome measures, in the late 1990s. With that idea as a foundation, they developed the AI fitting assistant FOX (Fitting to Outcome eXpert), which uses speech perception and other patient outcome tests as an input to its fitting algorithm. It analyzes a patient’s test results and previous MAPs against anonymized MAPs from other patients who were fitted with FOX to recommend the best possible MAP for that patient. The Belgian researchers validated the efficacy of FOX in a study, in which more than half of the 22 psychoacoustic targets, including those in areas like spectral discrimination and speech audiometry, were already achieved upon initial programming using the FOX method and the number of targets achieved increased by 24 percent once the FOX system was applied.3 Another study by them showed that the word recognition scores for one subject fitted with FOX improved at soft and loud intensities with the AI suggestions.4

Susan Waltzman, PhD, the Marica F. Vilcek Professor of Otolaryngology in the department of otolaryngology-head and neck surgery at NYU Langone Medical Center and the co-director of the NYU Cochlear Implant Center, further validated the viability of using FOX to program CIs in a study sponsored by Cochlear Americas, where she compared the speech perception performance in CI users who were fitted by an experienced clinician and standard programming methods versus the FOX-based programming algorithm.5 She found that the performance of a majority of those fitted with FOX was equivalent to that of those fitted by the traditional method, and a majority of the patients in the FOX group showed stable performance after only one month of experience with the new program. Eighty-five percent felt that the sound quality delivered through the system was excellent or good, and 45 of the 55 subjects said they would choose to continue with the FOX program.

READ  Germany seeks European cooperation on artificial intelligence

The high level of satisfaction with the FOX system, which is owned by Cochlear Americas, observed in the study can be attributed to the additional benefits it offers besides better speech perception outcomes. Dr. Waltzman said it’s advantageous to incorporate artificial intelligence, which is based on data, into CI programming. “It increases efficiency, allows for fewer visits to the implant center which makes it more convenient for the patient,” she said. “It also allows for the less experienced clinician to feel more confident that she/he is providing an appropriate program to the patient.” FOX recommends that newly implanted patients be seen four times a year for one to one and a half hours per session, including evaluations, greatly reducing the number of in-office visits CI recipients have to make to program their devices.

The lack of standardization in CI programming techniques is where a major potential of FOX really lies. A global survey of the current practices in CI fitting across 47 centers in 17 countries revealed that huge variation exists in virtually all aspects of CI fitting and follow-up, with each center having its own policy in timing, content, and methodology.6 It also found that the centers included relied mostly on the recipient’s subjective feedback to drive the MAP changes, and MAP parameters other than evoked potentials were rarely modified. “Over the past 30-plus years, there have been many changes to cochlear implants, including electrode arrays, processors, coding strategies, and other features, but programming methods have not changed substantially,” Dr. Waltzman said. “There is a lot of subjectivity involved.”

READ  CES 2019: Nvidia partners with Mercedes on artificial intelligence

Optimization of patient outcomes fundamentally requires results and targets that are measurable in most clinical settings and reliably repeatable, and that’s where FOX comes in. The setting of well-defined outcome targets through FOX allowed a range of different centers to apply a systematic methodology to monitoring the quality of the CI programming provided in one study.3

The use of clear targets enables audiologists to define what is meant by an optimized MAP, and provides consistency across different professionals and centers. This is especially helpful for patients who move to another geographic area and need to switch CI centers. There is comfort knowing that the level of care or performance will not be compromised or affected by individual approaches to programming.

Dr. Waltzman emphasized in her study that FOX should not be construed to be a substitute for an experienced clinician but rather a tool an audiologist can use to provide patients with the best possibility for success through a more standardized evidence-based approach. “When the software becomes available, audiologists who program cochlear implants should explore incorporating using the AI technique into their repertoire,” she said. “While it does require changing a mindset and methodology, it could enhance services and increase efficiency and could prove to be a valuable tool for the audiologist to use at their discretion.”

REFERENCES

1. Davenport, T, and Kalakota, R. The potential for artificial intelligence in healthcare. Future Healthc J. 2019 Jun;6(2):94-98.


3. Vaerenberg, et al. Setting and Reaching Targets with Computer-Assisted Cochlear Implant Fitting. ScientificWorldJournal. 2014;2014:646590.

4. Wathour, et al. From Manual to Artificial Intelligence Fitting: Two Cochlear Implant Case Studies. Cochlear Implants Int. 2019 Sep 17:1-7. doi: 10.1080/14670100.2019.1667574.
READ  What Does Elon Musk Think of AI?

5. Waltzman, S, and Kelsall, D. The Use of Artificial Intelligence to Program Cochlear Implants. Otol Neurotol. 2020 Apr;41[4]:452-457. doi: 10.1097/MAO.0000000000002566.

6. Vaerenberg, et al. Cochlear Implant Programming: A Global Survey on the State of the Art. ScientificWorldJournal. 2014 Feb 4;2014:501738.



READ SOURCE

LEAVE A REPLY

Please enter your comment!
Please enter your name here