Otoacoustic Emissions
Design, run, and score distortion-product OAE (DPOAE) testing — read the DP-gram, call a category at each frequency, and issue a pass/refer per ear.
The Otoacoustic Emissions element simulates distortion-product OAE (DPOAE) testing. For each ear it presents a set of test frequencies; at each one the learner acquires the measured DP amplitude and noise floor, reads the DP-gram, calls a category (present, present but low amplitude, absent, or noisy), and issues an overall pass / refer decision per ear. Those two answers — the category at each frequency and the ear's overall call — are the whole task. In the designer it's listed as:
Distortion product OAEs — read the DP-gram, call each frequency, and pass/refer
Like every Theta element, it follows the lifecycle Design → Settings → Testing → Scoring → Report.
Three criteria, one category
Three criteria decide each frequency, and the noise floor is applied first, because it decides whether the other two can be believed at all:
- Noise floor — NF ≤ the maximum. The recording is clean enough to read.
- SNR — DP − noise floor ≥ the criterion. An emission is there at all.
- Amplitude — DP ≥ the minimum DP level. It is big enough to be a normal-range emission.
They produce exactly one of four categories:
| Noise floor | SNR | Amplitude | Category |
|---|---|---|---|
| Too high | (either) | (either) | Noisy — the frequency was not measured; no call can be trusted |
| OK | Not met | (either) | OAE absent |
| OK | Met | Not met | OAE present, low amplitude — a real but weak emission; doesn't count toward a pass |
| OK | Met | Met | OAE present — counts toward a pass |
A noisy floor gates a present call too, not just an absent one. That is what makes Noisy mean one thing: the recording was too dirty to conclude anything there. The learner reports what the OAE actually showed — there is no re-run button. If you want them to see a clean retest, add a second OAE element.
An ear Passes when the pass rule's required number of frequencies are both readable and present. A noisy frequency is never counted as absent — that would claim an emission is missing on a recording that cannot show one — so an ear without enough readable frequencies left simply cannot reach the bar and Refers: an inadequate recording is not a pass. There is no "inconclusive" ear.
Simplified for now
The underlying SNR OK?, Amplitude OK? and Noise floor OK? reads still exist in the model and the scorer, but the learner is not asked for them and they are not scored. The task is deliberately two answers — the category and the overall call — and the component reads can be brought back as separately taught and scored steps later.
Design
The Design tab is where you set the true emission picture the learner must discover, per ear and per frequency.
Protocol and frequencies
Two starting protocols are offered:
| Protocol | Frequencies (Hz) |
|---|---|
| Screening | 2000, 3000, 4000, 5000 |
| Diagnostic | 500, 1000, 2000, 3000, 4000, 6000, 8000 |
You can add or remove any f₂ frequency from the master list (500 – 8000 Hz); at least one is required.
Per-frequency values: template vs fixed
As with the other clinical elements, template vs fixed is chosen on the element card.
| Mode | What you set per frequency |
|---|---|
| Fixed values | The exact DP amplitude and noise floor (dB SPL) used on every submission. |
| Template | One category per frequency: Present, Present · low amp, Absent or Noisy. Each submission draws realistic dB values inside that category, so numbers vary but the picture stays stable. |
Present generates a DP that clears both the SNR criterion and the minimum DP level on a clean floor; Present · low amp generates a clearly detected DP that sits under the minimum level (a mild-loss picture); Absent generates a DP sitting in the noise on a clean floor; Noisy generates a floor above the maximum, with the DP lost in it. The category sets the noise floor it implies, so a row can never claim a picture its numbers do not support.
In fixed mode the dB numbers are the design — editing a DP or NF value re-derives that row's emission and noise-floor category under the current criteria. The category you see always matches the numbers the learner will see, so the answer key can't drift away from the case.
Quick-fill patterns
To seed both ears fast (then edit individual cells), pick a pattern — All present (normal), Present, low amplitude (mild loss) (every frequency detected but under the minimum level), All absent (refer), Partial (high-frequency loss) (frequencies ≥ 3 kHz absent), or Noisy recording (elevated noise floor, with some emissions poking through and the rest buried) — and apply it to the Right ear, Left ear, or bilaterally.
A quick-fill only rebuilds ears that exist: an ear you removed with Remove ear stays untested. "Affected ear" chooses which ear is abnormal, not which ears are present.
Batches — OAE screening
Add a batch… → OAE screening drops a set of screening patients into the lesson, dealt fresh every attempt. You set how many, the refer rate (the dialog shows what that deals — exactly 2 refer · 8 pass at 20% of 10), and the protocol:
- Screen at — tap the f2 frequencies every patient is screened at (2, 3, 4 and 5 kHz to start). At least one stays on.
- Pass when at least N … are present — the pass rule, starting from your preset's: a frequency counts only when its emission is Present (an emission too weak to count — Present · low amp — doesn't). It can't exceed the number of frequencies screened.
The protocol applies to every patient in the batch, including ones already added. Then:
- A pass has clear emissions in both ears. A refer has no emissions, emissions too weak to count (present but under the amplitude floor), or loses them from 3 kHz up — in one or both ears. Every patient is checked against the pass rule and re-drawn if it lands on the wrong side (a 3 kHz drop-out on a protocol that screens nothing that high becomes a full absence).
- The learner reads each DP-gram and calls each ear Pass or Refer — the overall result only. The per-frequency category row isn't shown, and nothing per frequency is graded.
- The results screen and PDF open with a batch summary: overall accuracy, sensitivity (refers caught), specificity (passes left alone), unanswered ears, and one line per patient.
The same rule applies to any OAE element: the category row appears only when the category is scored or report-only.
A batch lesson holds only its batch and Information elements — see Audiometry → Batches for how batches work.
Settings
The Settings tab holds the element-wide validity rules and display values. The interpretation rules update live as you change them.
| Setting | Default | What it does |
|---|---|---|
| SNR criterion | 6 dB | An emission is detected when DP − noise floor ≥ this. |
| Minimum DP level | −5 dB SPL | A detected emission below this is present, low amplitude rather than present. |
| Max trusted noise floor | 10 dB SPL | A frequency whose noise floor exceeds this is Noisy — not measurable, whatever the SNR and amplitude say. Also shapes the numbers generated for that category. |
| Min frequencies to pass | 3 | Overall pass needs at least this many tested frequencies present (both reads OK). |
| L1 / L2 | 65 / 55 dB SPL | Primary tone levels, shown for context. |
| f2/f1 ratio | 1.22 | Primary frequency ratio, shown for context. |
| Plot y-axis | −25 to 30 dB SPL | DP-gram amplitude bounds. |
Where the defaults come from
6 dB SNR is the near-universal screening criterion; a minimum DP level of −5 dB SPL is what most screeners ship with (adult diagnostic norms use −10; infant protocols 0 to +5). Both are editable here and in an organization's element preset.
These are the values an organization's settings preset seeds into every new OAE element. The preset holds the criteria and stimulus values only — protocol and tested frequencies are per-case, so they're set in the element, not the preset. A seeded element's dB values are built against that preset's criteria, so a preset with, say, a 10 dB SNR criterion starts you with a DP-gram that genuinely reads as present under 10 dB.
Changing a criterion re-aligns the design
The SNR criterion, minimum DP level and max trusted noise floor are the boundaries every stored point sits against, so changing one updates the design to match — in whichever direction that element's mode makes correct:
- Fixed — the dB numbers are the design and the answer key is derived from them, so the categories are recalculated. Raise the SNR criterion past what a row clears and that row's category changes to Absent in front of you, rather than continuing to read Present while the learner is marked against Absent. Lower the max trusted noise floor under a row's floor and it becomes Noisy.
- Template — the categories are the design and the numbers are only a preview, so the numbers are redrawn inside the categories you set. Shuffle (in the template banner) draws a new preview without changing a category. The case you authored survives the change, and every attempt keeps generating from it.
The pass rule, primary levels and plot bounds don't decide any row's reads, so they leave the design untouched.
Testing
Each ear gets its own panel (they stack on small screens). A context bar shows the stimulus settings and the present criterion (e.g. SNR ≥ 6 dB).
The learner:
- Acquire sweeps the ear chosen in the console (switch ears in the tray to test the other). The ear's noise floor appears first, at every frequency, flickering as it settles; then each frequency's primary tones play (with a faint distortion product where an emission is detected) and its DP lands on the floor, in the DP-gram and in the table, with its SNR (DP − NF). The floor holds still at its measured values once the sweep ends. Once a sweep has finished the button becomes Re-acquire, so the run can be repeated the way a real device would be; it replays the same measurement, it doesn't redraw it.
- Category — Present, Present · low amp, Absent, or Noisy — when the case asks for it (the category is Scored or Report-only; Off removes the row).
- Record the overall determination for the ear — Pass or Refer.
In practice mode an optional SNR guide can be overlaid on the plot and an answer key revealed. Both are study aids, so in exam mode neither is offered.
Every step is recorded to the attempt's action trail — acquisitions (with the values revealed), each category call and the overall call — so an instructor can reconstruct how the learner worked.
Scoring
Scoring is a tri-state per field, per ear — each field is Scored, Report-only (shown on the report, not counted), or Off. The Right and Left ears configure independently, with a Sync Ears toggle to mirror them.
| Field | What's compared |
|---|---|
| Category (per frequency) | Learner's Present / Present · low amp / Absent / Noisy vs the call derived from the design. Off also removes the category row from the learner's console — the case then asks only Pass / Refer; Report-only keeps it asked but ungraded. |
| Overall pass/refer | Learner's decision vs the pass rule (min frequencies to pass both readable and present, across every tested frequency). |
Comparisons are exact — the criteria are pass/fail rules, so there's no tolerance band. An unanswered call counts as wrong. Advanced Analytics carries Emission call accuracy and Pass / refer determination so a cohort report can show where learners are going wrong.
Which frequencies are scored
Every frequency the element requires (the Settings tab's frequency list) earns per-frequency marks for an ear that counts — change the tested set and the scored set follows it. To leave a frequency out of the score, leave it out of the test.
Older attempts keep their grade
Until Theta began scoring every required frequency, a lesson could leave frequencies out of the per-frequency score — on the Design tab, and later on the Scoring tab. An attempt started before that change keeps the frequencies it was graded on, so its score doesn't change when its results, report or submissions list are opened, or when it is re-scored. Every attempt started since scores every required frequency.
A lesson that still stores such a list says so on its Scoring tab: an amber note names each ear and the frequencies it used to leave out. To leave one out of the score now, remove it from the test on the Settings tab.
The ear's overall Pass/Refer is a separate field with its own Scored / Report-only / Off control.
An ear the design doesn't test is shown unchecked and locked, with a note to add it on the Design tab; add it back and its scoring comes on by default.
Teaching with the Noisy category
Build a case with the Noisy recording pattern when the teaching point is measurement quality. The learner should recognise that the floor is too high to conclude anything at that frequency and call it Noisy — which means "re-seat the probe and re-run", not "no emission". Because a noisy frequency is never counted as absent, an ear with too few readable frequencies left cannot reach the pass rule and refers: an inadequate recording is not a pass.
Report
After submitting, the results screen carries an OAE breakdown (every scored and report-only field, expected vs submitted) and a View data tab with the DP-gram and a per-frequency comparison.
On the PDF, each configured ear gets its DP-gram — green markers for present and amber for present · low amp, over the shaded noise-floor band — above a table.
Like the designer's own preview, this is an answer-key view, so it marks only the frequencies where an emission was detected. Absent and noisy ones carry no marker and the trace breaks around them, because there is no response to plot; which of the two it is reads off the noise floor beneath — absent sits over a clean floor, noisy over a raised one. The learner's plot is different: it marks every measured DP, because deciding whether one clears the criterion is exactly the judgement being taught.
The table beneath:
| Column | Content |
|---|---|
| Freq | Test frequency (f₂). |
| DP/NF/SNR | The measured values and their signal-to-noise ratio (dB SPL). |
| Expected | The correct reads — e.g. SNR OK · Amp not OK · Present · low amp (with NF OK / not OK added when that read is on). |
| Submitted | What the learner recorded, with — for any read left blank. |
| Result | ✓ when every scored read on the row is right, ✗ otherwise, — for report-only. |
A final Overall row compares the expected pass/refer against the learner's decision. A frequency is listed only where a read or the category was asked (Scored or Report-only), here and on the results screen: an ear that asks only Pass / Refer, such as an OAE screening, shows its DP-gram and the Overall row alone. If the instructor set every field to Off, the element is out of the record entirely and no block is printed — and an individual ear unticked on the Scoring & Report tab is left out the same way, on both the PDF and the results screen. Off means out of the record; use Report-only for an ear you want shown but not marked.