Comparing AutoREM Accuracy to NAL-NL2 Targets: Real Patient Data and Simulator Verification

Emily Brand- Audiologist. Click Hearing- Essex, UK

Automatic Real Ear Measurements (AutoREMs) have emerged as a promising way to make hearing aid fittings faster and more consistent. But do they really deliver the accuracy that clinicians expect from traditional Real Ear Measurements (REMs)? This feature explores a new study that compared AutoREM performance from four leading manufacturers - Phonak, Signia, Starkey, and Oticon - using both simulator and real patient data. The results shed light on how well AutoREMs align with NAL-NL2 prescription targets and what this means for everyday clinical practice.

A quick refresher: What are AutoREMs?

REMs remain the gold standard for verifying that a hearing aid delivers the prescribed level of amplification into a patient’s ear canal. However, despite their proven value, many clinicians skip this step due to time constraints, lack of confidence, or a belief that manufacturer defaults are ‘good enough’. This can lead to mismatches between target and output, potentially compromising speech understanding, listening comfort, and overall satisfaction.

AutoREMs aim to tackle this challenge by automating the measurement and adjustment process. They integrate probe microphone measurements directly into the manufacturer’s fitting software, allowing the software to automatically adjust the hearing aid’s settings to match the prescription target based on real-time measurements in the patient’s ear canal. Theoretically, this streamlines the workflow - but there’s a key question: do these automatic fits truly hit the mark?

The study in focus

This study examined how accurately AutoREMs from four major brands could match NAL-NL2 targets. Measurements were taken at 50, 65, and 80 dB SPL using the International Speech Test Signal (ISTS). To reflect both controlled and real-world conditions, the researchers combined data from AHead Solutions’ CARL ear simulator with real patient measurements.  

In total, 414 datasets were collected: 72 per manufacturer from the simulator and around 30 per manufacturer from clinical patients. Models tested included Phonak (Infinio Sphere 90, Naida Lumity 90), Signia (Pure 7IX, Motion 7X), Starkey (Edge AI 24 RIC, Evolv 24 BTE), and Oticon (Intent 1 RIC, Real 1 BTE). Various hearing loss configurations were represented - from sloping high-frequency sensorineural losses to flat and upward-sloping profiles. All fittings used the NAL-NL2 prescription and were verified using the Natus Freefit system.

What the data showed

Across manufacturers, AutoREMs generally stayed within the ±5 dB tolerance recommended by the British Society of Audiology, indicating that most systems delivered accurate results. However, performance varied depending on frequency, input level, and manufacturer.

Signia demonstrated the most consistent accuracy across all input levels and frequencies. Phonak tended to deviate more at low frequencies (especially at 50 dB SPL), while Oticon showed greater variability at higher frequencies. Starkey’s results fluctuated more at lower inputs but stabilised at conversational and loud speech levels.

Notably, Phonak was the only manufacturer to exceed the BSA's recommended 5 dB tolerance threshold at 50 dB SPL (RMSE = 5.67), though the difference wasn’t statistically significant (p = 0.26). In practical terms, these findings suggest that while AutoREMs perform well overall, accuracy can drop at lower input levels - an important consideration for softer speech or quiet environments.

Figure 1 shows that outliers highlight extreme deviations, particularly at higher frequencies such as 5000 Hz and 6300 Hz, suggesting abnormal values that deviate significantly. Clinicians should therefore be cautious when interpreting threshold differences at higher frequencies. Real-world variables, such as patient movement, ear canal variability, and differences among manufacturers’ algorithms, can all introduce discrepancies that AutoREM systems might not fully account for.

Figure 1: All Manufacturers Data Boxplot

Table 1: All Manufacturers Data Stats

Figure 2: All manufacturers’ data averages per frequency

Making sense of the results

The combination of both simulated and real patient data strengthens these findings. Simulator testing provided consistency and control, while clinical data captured real-world variability - such as ear canal differences, venting effects, and patient movement. Together, these datasets paint a realistic picture of how AutoREMs behave in daily practice.

AutoREMs are impressively close to manual verification in many cases, particularly at conversational and louder levels. However, small discrepancies can still occur, especially in soft speech conditions or at specific frequency regions. For clinicians, this means AutoREMs are a valuable time-saving tool - but they don’t replace the need for professional verification and adjustment when precision matters most.

Implications for clinical practice

Clinicians increasingly face pressure to deliver accurate, efficient fittings in less time. AutoREMs can help bridge that gap, offering a quick check that fittings are within tolerance. However, as with any automated process, they should be used with clinical judgement. Experienced audiologists will still want to verify complex fittings manually - for example, in cases involving unusual ear canal acoustics, severe losses, or mixed hearing configurations.

Used appropriately, AutoREMs can enhance workflow, free up clinical time, and improve consistency across fittings. They may also help newer clinicians develop confidence with verification techniques before moving to full manual REMs. In short, AutoREMs are not a substitute for expertise - but a practical tool that complements it. 

AutoREMs continue to evolve, and manufacturers are clearly closing the gap between automated fitting approaches and manual verification. Future research should expand testing across more hearing aid styles, different degrees of hearing loss, and a wider range of clinical conditions. Refinements at lower input levels would further boost confidence in AutoREM accuracy, ensuring patients receive optimal benefit from their fittings regardless of environment or device brand. As technology advances, AutoREMs will likely become a bigger part of audiology practice, but clinician insight will always be essential for achieving the best individual outcomes.

References:
Brockmeyer, A., Voss, A., Wick, C.C., Durakovic, N. and Valente, M. (2021). Accuracy of an Automated Hearing Aid Fitting Using Real Ear Measures Embedded in a Manufacturer Fitting Software. Journal of the American Academy of Audiology, 32(03), pp.157–163. doi:https://doi.org/10.1055/s-0041-1722947.

BSA (2018). Practice Guidance Guidance on the verification of hearing devices using probe microphone measurements. [online] Available at: https://www.baaudiology.org/app/uploads/2020/11/REMS-2018.pdf.

Denys, S., Latzel, M., Francart, T. and Wouters, J. (2018). A preliminary investigation into hearing aid fitting based on automated real-ear measurements integrated in the fitting software: test–retest reliability, matching accuracy and perceptual outcomes. International Journal of Audiology, 58(3), pp.132–140. doi:https://doi.org/10.1080/14992027.2018.1543958.

Mueller, G. and Ricketts, T. (2018). 20Q: Hearing Aid Verification - Will AutoREMfit Move the Sticks? [online] AudiologyOnline. Available at: https://www.audiologyonline.com/articles/20q-hearing-aid-verification-226-23532.

Acknowledgements:

Ben Mann

Phonak, Signia, Starkey, Oticon

Guymark (AHead Simulations)
Declaration of competing interests: None declared.