Where Does ReFittingLab Fit
in Your Clinical Workflow?
Current best practice excels at measuring what reaches the ear — REM, REAR, audiogram targets. But it has no standardised way to measure what the patient actually perceives and discriminates after fitting. That's the gap ReFittingLab fills — with a patent-backed, evidence-based algorithm that includes 5 independent safety layers and a multi-factor computation engine.
The Fine-Tuning Gap
After applying a prescriptive formula and verifying with REM, the clinician enters the fine-tuning phase — and suddenly loses access to objective data.
❌ What happens today during fine-tuning
- "How does it sound?" — Clinicians ask subjective questions with no quantifiable answers
- Trial and error — Adjust gains, send patient home, wait for next appointment
- 3–5 return visits — Each one essentially repeating the same subjective loop
- No phoneme-level data — If the patient says "I can't understand speech," the clinician doesn't know which frequencies are causing the issue
- Patient frustration — 30% of hearing aid users abandon their devices within 12 months
✅ What changes with ReFittingLab
- Objective phoneme discrimination scores — Per-contrast, per-frequency, quantified across 3 intensity levels
- Targeted adjustments — The multi-factor algorithm tells you exactly which bands need adjustment, calibrated to the patient's audiometric profile
- 5 safety validations — UCL proximity, loudness summation, dynamic range, MPO headroom, and compression ratio are checked before any recommendation
- Fewer visits — More gets resolved in the first session because you have data, not guesses
- Clinical documentation — Every decision is evidence-based, auditable, and IEC 62304 traceable
The verification gap in current practice
Pre-fitting has excellent objective tools: audiogram, REM/REAR, RECD, prescriptive formulas (NAL-NL2, DSLv5.0, manufacturer-based).
Post-fitting fine-tuning has almost none. The clinician adjusts based on subjective patient reports and their own experience — with no objective way to measure whether word understanding actually improved after a gain change. Did the patient's ability to distinguish similar-sounding words get better? Worse? By how much? Nobody knows.
ReFittingLab bridges this gap with a complete fine-tuning verification system: it evaluates the patient's ability to discriminate and understand speech sounds — using minimal pairs, word lists, and speech-in-noise tests — then maps the results directly to per-band gain adjustments. The entire process is backed by a multi-factor computation engine with 5 independent safety layers, producing targeted, bounded recommendations that improve real-world word understanding, not just acoustic output.
Where ReFittingLab Fits in Your Workflow
ReFittingLab does not replace your fitting software or prescriptive formula. It enters the workflow after the initial fitting, exactly where objective data disappears.
❌ Without ReFittingLab
- Fitting session ends after REM verification — no measurement of actual word understanding
- Patient goes home with "come back in a few days and tell me how it sounds"
- If unhappy, clinician adjusts gains based on vague subjective feedback with no data on which sounds are problematic
- 3–5 return visits, each repeating the same subjective loop — costly for the clinic, frustrating for the patient
- No documentation of fine-tuning rationale — if audited, no evidence trail of why adjustments were made
✅ With ReFittingLab
- After REM, run a 20-minute verification session — objective measurement of the patient's actual word discrimination
- Algorithm identifies exactly which frequency bands need adjustment and by how much — validated against 5 safety checks
- Apply the recommendations in your fitting software, re-test, and confirm measurable improvement — all in the same session
- Patient leaves with a data-verified fitting — fewer return visits, higher satisfaction, lower abandonment
- Full audit trail: every test, every recommendation, every re-test is documented and IEC 62304 traceable
Audiogram & Patient History
Standard audiometric evaluation: air/bone conduction, UCL, speech recognition threshold. This is your baseline.
Prescriptive Formula
Apply your formula of choice — NAL-NL2, DSLv5.0, or a manufacturer-based formula (Phonak APD, Oticon VAC+, etc.) — to generate initial gain targets.
Initial Fitting in Manufacturer Software
Program the hearing aids in Phonak Target, Oticon Genie 2, ReSound Smart Fit, Signia Connexx, Widex Compass GPS, or any other fitting software. Configure coupling, feedback management, features.
Real-Ear Measurement (REM/REAR)
Verify that the acoustic output at the eardrum matches prescriptive targets. This confirms the physical delivery of sound — but not the perceptual outcome.
ReFittingLab — Phoneme Discrimination Verification
This is where ReFittingLab enters. Using extremely select minimal pairs — curated through rigorous phonological, phonetic, and psychometric criteria — the system evaluates the patient's ability to discriminate speech sounds across frequency bands at multiple intensity levels (50, 65, 80 dB SPL), and outputs specific, per-band gain adjustments (ΔG50, ΔG65, ΔG80) validated through 5 independent safety checks.
The clinician then applies these adjustments in their fitting software and re-tests to confirm measurable improvement. No guesswork. No waiting for the next appointment.
Fine-Tuning with Data
Apply ReFittingLab's recommendations in your fitting software. Adjust G65 for conversational speech, G50 for soft sounds, G80 for loud environments — each adjustment mapped to a specific phoneme discrimination deficit and validated against the patient's dynamic range.
Verified Outcome & Discharge
Re-test with ReFittingLab to confirm improvement. Document results with statistical significance verification (Thornton-Raffin critical difference). Discharge with confidence — or schedule follow-up only if objective data warrants it.
What Happens During a ReFittingLab Session
A complete verification session takes 20–30 minutes and produces actionable, per-band gain adjustment recommendations — each validated against 5 safety layers.
Input Patient Profile
Enter the audiogram (air/bone conduction, UCL), current gain tables (G50/G65/G80/CR/MPO from your fitting software), and clinical context: experience level, tinnitus, hyperacusis, coupling type, bilateral status.
Automatic Test Selection
Based on the audiometric profile, the system selects which phoneme contrasts to prioritise. Steep high-frequency loss? Prioritise fricatives (/s/ vs /θ/). Flat moderate loss? Include stops and nasals. Severe HF loss? Flag for frequency lowering verification.
Phoneme Discrimination Testing
Calibrated minimal pair stimuli — each pair differing in exactly one phonemic segment — are presented via a forced-choice paradigm. The patient selects what they hear: each error maps to a specific frequency band through the distinctive feature → frequency correspondence table.
Multi-Level Testing
Extend to 50 dB (audibility of soft sounds) and 80 dB (comfort and tolerance). This maps to G50, G65, and G80 adjustments across the full dynamic range — not just conversational level.
Speech-in-Noise Verification
Sentences are presented with calibrated noise (ICRA/ISTS) at multiple signal-to-noise ratios (+5, 0, −5 dB). Calculates the patient's SNR loss — how many extra dB of signal they need compared to a normal listener.
Multi-Factor Analysis & Safety Validation
The algorithm maps each phoneme error to its acoustic correlate (distinctive feature → frequency band → NFM band matching), then computes the gain delta using a multi-factor formula that considers hearing threshold, dynamic range, session history, and effective audibility. Every recommendation passes through 5 independent safety checks before presentation.
Apply in Your Fitting Software
Take the recommendations and apply them directly in Phonak Target, Oticon Genie 2, ReSound Smart Fit, or whichever fitting software you use. ReFittingLab tells you what to change, why, and how much is safe — you stay in control.
Re-Test & Verify
Re-administer the tests that showed deficits. If the adjustment was correct, discrimination scores improve measurably. Statistical significance is verified using Thornton-Raffin critical difference tables, with full longitudinal tracking across sessions.
From Phoneme Error to Safe Gain Adjustment
The core algorithm translates a perceptual error into a precise, band-specific gain change through a multi-stage pipeline. Each adjustment is bounded, safe, and grounded in the patient's individual audiometric profile.
Speech Error
"thin" heard as "sin"
Distinctive Feature
[+strident] vs [−strident]
NFM Band Mapping
→ 4 kHz band
Multi-Factor ΔdB
base × EA × DR × session
5 Safety Checks
UCL · MPO · LS · DR · CR
This is not a generic "boost highs if speech is unclear" approach. Each phoneme contrast maps to a specific set of distinctive features (Jakobson-Halle-Fant framework), which in turn correspond to specific frequency bands via the Nearest Frequency Matching (NFM) mechanism — automatically adapting to any hearing device's band configuration (5-band consumer earbud or 20-band clinical aid). The multi-factor formula considers:
Effective Audibility (EA)
Based on Ching, Dillon et al. (2001, NAL-NL2): as hearing loss increases, the ability to extract speech information diminishes even when sound is audible. The algorithm scales adjustments per-band based on the patient's threshold — not a naive flat boost. At 66 dB HL, contribution drops ~50%.
SII-Weighted Prioritisation
Band importance follows ANSI S3.5-1997 Speech Intelligibility Index weights. The 2–4 kHz region contributes ~54% of speech intelligibility — adjustments in these bands are computed and prioritised proportionally.
Session-Aware Graduation
First-session adjustments are conservatively bounded. Subsequent sessions can build on previous data, with the algorithm tracking longitudinal improvement per contrast per band — enabling precise, progressive optimisation.
Dynamic Range Awareness
The gain delta is bounded by each band's available dynamic range (UCL − threshold). Patients with narrow dynamic range (< 30 dB) receive proportionally smaller adjustments to preserve comfort headroom.
NFM — Nearest Frequency Matching
The core technology that makes ReFittingLab universally compatible: NFM automatically maps each speech-relevant frequency to the closest adjustable band of any hearing device — whether it has 4 bands (basic aid), 10 bands (mid-range), or 20 bands (premium clinical device). The system consolidates multiple speech frequency regions that fall within the resolution of a single band, weighting each by the magnitude of the discrimination deficit and spectral proximity. One evaluation → any device → optimal result regardless of band count.
Cross-Device Portability
Discrimination data is device-independent — it characterises the patient's perceptual profile, not the hearing device. When the patient changes devices, only steps (b)–(d) are re-executed with the new band configuration. No re-evaluation needed.
5-Layer Clinical Safety Validation
Every adjustment recommendation passes through 5 independent safety checks before being presented to the clinician. This isn't just gain boosting — it's a validated, bounded computation with multiple protective mechanisms.
1. UCL Proximity Check
Estimates the post-adjustment output level and verifies it stays at least 7 dB below the patient's Uncomfortable Loudness Level (ASHA guideline: 5–10 dB margin). If exceeded, the system caps the recommendation and flags the band.
2. Spectral Loudness Summation
When 3 or more high-frequency bands receive simultaneous gain increases, perceived loudness exceeds their arithmetic sum (Keidser & Moore). The system applies a 1.5× summation factor and limits total perceived HF increase to 8 dB per session.
3. Dynamic Range Protection
For patients with narrow dynamic range (< 30 dB between threshold and UCL), adjustments exceeding 3 dB are flagged with specific warnings. The recommendation is bounded to preserve comfortable headroom.
4. MPO Headroom Verification
Cross-references the proposed adjustment against the hearing device's Maximum Power Output to ensure the recommendation doesn't push the output level into saturation. Requires the actual gain table from fitting software.
5. Compression Ratio Guard
For frequency bands involved in temporal contrasts (voiced/voiceless stop consonants: /p/ vs /b/, /t/ vs /d/, /k/ vs /g/), the system detects compression ratios exceeding 3:1 — which degrade Voice Onset Time discrimination (up to 19.4% intelligibility loss).
All safety constants are derived from published clinical guidelines (ASHA, BSA, NAL-NL2, ANSI S3.5-1997, Keidser & Moore) and are configurable per-deployment. The system generates a ValidationReport classifying the overall recommendation as PROCEED, PROCEED WITH CAUTION, or REVIEW REQUIRED.
Works With Your Existing Fitting Software
ReFittingLab is manufacturer-agnostic. It generates per-band gain adjustment recommendations that you apply in whichever fitting software you already use. The NFM mechanism automatically adapts to any device’s band configuration — from basic hearing aids with 4 bands to premium devices with 20 or more.
Phonak Target
Paradise, Lumity, Infinio
Oticon Genie 2
More, Real, Intent
ReSound Smart Fit
ONE, Nexia, Omnia
Signia Connexx
IX, AX, Pure
Widex Compass GPS
SmartRIC, Moment, Sheer
Any Other Software
With adjustable gain bands
All product names mentioned are trademarks of their respective owners. Novaural is not affiliated with, endorsed by, or sponsored by any hearing aid manufacturer. ReFittingLab is an independent verification system.
What You Need
ReFittingLab requires no booth modifications and no proprietary hardware. The entire setup fits on your existing clinic desk — a laptop, a speaker, and a sound level meter. That's it.
Active Studio Monitor
A powered near-field monitor with flat frequency response across the speech range (200 Hz – 8 kHz). Studio reference monitors designed for accurate audio reproduction are ideal — widely available and affordable.
~£100 · Lasts yearsSound Level Meter (Class 2)
An IEC 61672-1 Class 2 compliant sound level meter. Used only during the initial calibration to verify SPL at the patient position. Once calibrated, it stays in the drawer until you change rooms or equipment.
~£100 · One-time use per setupReFittingLab Software
Runs on any Windows or Mac laptop. Includes a built-in calibration wizard, language packs, and the full multi-factor algorithm with all 5 safety layers. No proprietary dongles, no per-session limits.
Early adopter rates availableOne-Time Calibration — 5 Minutes
The software includes a guided calibration wizard that takes under 5 minutes:
- Place the sound level meter at the patient position (typically 1 metre from the speaker)
- The software plays calibrated reference tones through the loudspeaker
- Read the SPL from the meter and enter it — the software auto-adjusts output to match the target levels (50, 65, 80 dB SPL)
- Calibration is stored — only needs re-doing if you change the room, speaker, or position
The process follows IEC 60645-1 calibration principles. No specialist technician required — any audiologist can complete it independently.
Total hardware investment: under £250
Less than the cost of a single hearing aid return. Equipment lasts thousands of sessions with no consumables. If you already own a suitable speaker and sound level meter, your cost is zero.
Not sure which hardware to choose? We're happy to advise on the minimum specification that meets clinical requirements. We are not affiliated with, sponsored by, or receiving commissions from any hardware manufacturer — we only recommend what is sufficient for reliable, standards-compliant sessions.
Better Outcomes for Everyone
ReFittingLab creates measurable value across the entire hearing care ecosystem — from NHS trusts managing thousands of patients to private practices competing on quality.
For NHS & Public Health
Efficiency at scale
- ✓ Reduce follow-up visits — Objective verification in the first session means fewer return appointments clogging waiting lists.
- ✓ Shorten waiting lists — Faster, more efficient fitting sessions free up clinic capacity for new patients.
- ✓ Standardise quality — Quantitative protocols reduce variability between clinicians and clinics across trusts.
- ✓ Evidence for commissioners — Measurable outcomes data supports service quality reporting and funding justification (CQC audit-ready).
For Private Clinics
Competitive advantage
- ✓ Differentiating technology — Offer patients verifiable proof of fitting quality that competitors cannot match.
- ✓ Reduce staff costs — Shorter sessions and fewer follow-ups mean lower cost per patient and higher throughput.
- ✓ Higher conversion rates — Patients who experience measurable improvement in-session are more likely to commit to purchase.
- ✓ Lower return rates — Getting the fitting right the first time reduces costly device returns and exchanges.
Why Objective Fine-Tuning Data Matters
From Subjective to Objective
Instead of "How does it sound?", you get "/s/ vs /θ/ discrimination: 68% → needs +1.19 dB at 4 kHz (bounded by EA factor 0.85, DR factor 0.70)". Every adjustment has a reason. Every reason has data.
Targeted, Not Global
ReFittingLab doesn't suggest "boost highs." It identifies which frequency band is causing which phoneme confusion through distinctive feature analysis, and calculates the minimum effective adjustment — typically 0.5–3 dB, bounded by 5 safety layers.
Closed-Loop Verification
Adjust → Re-test → Confirm. In the same session. Statistical significance is verified using Thornton-Raffin critical difference tables. The system tracks longitudinal improvement per contrast per band across sessions.
Clinical Documentation
Every test result, every recommendation, every safety check, every re-test is recorded in an IEC 62304 traceable audit trail. Useful for clinical governance, audits (NHS/CQC), and longitudinal patient outcomes tracking.
More Done in Fewer Visits
Data-driven adjustments resolve more issues in the first fitting session. The traditional 3–5 visit cycle compresses — better outcomes, happier patients, more efficient clinics.
Language-Independent Architecture
The system uses interchangeable language packs with language-specific phoneme inventories, minimal pairs curated through rigorous phonological criteria, and fully localised clinical terminology. Currently calibrated for English, with the architecture designed to support any additional language — simply add a validated language pack and the entire system adapts automatically.
Ready to See It in Action?
We're seeking clinical partners to validate ReFittingLab in real-world settings. If you're an audiologist, hearing aid dispenser, or clinic manager, we'd love to hear from you.