FDA Codifies Four Device Types Into 21 CFR in a Single Day: OTC Sleep Apnea Risk Assessment, Skin Cancer Adjunctive Software and Two Imaging Systems

On 30 October 2026, the U.S. Food and Drug Administration published four final device classification orders in the Federal Register, formally writing into Title 21 of the Code of Federal Regulations four device types that had previously been classified through the De Novo pathway: 21 CFR 868.2378 (over-the-counter device to assess risk of sleep apnea), 878.1830 (software-aided adjunctive diagnostic device for physician use on lesions suspicious for skin cancer), 892.2085 (radiology software for referral of findings related to fibrotic lung disease) and 868.1985 (spinal imaging system for neuraxial procedures). All four are classified as Class II with special controls, and the orders take effect on the date of publication. The underlying classification decisions themselves have applied since 12 January, 19 January and 6 February 2024 respectively. The special controls prescribe, item by item, clinical study populations, subgroup analyses, comparator methods, independent validation sets for machine learning and labeling wording. Similar products will no longer need to go through De Novo and can enter the U.S. market via the 510(k) comparison route.

Four Final Orders Filed and Published Together

The four FDA final orders were docketed on 29 October 2026 and published in the Federal Register on 30 October 2026, with consecutive docket numbers: FDA-2026-N-12213, FDA-2026-N-12254, FDA-2026-N-12255 and FDA-2026-N-12385. All four take the same action: converting devices already classified through De Novo into codified classification regulations in Title 21 of the CFR.

These are not four unrelated notices. They point in one direction across four product categories: over-the-counter self-assessment, adjunctive skin imaging interpretation, lung imaging referral and puncture-localization imaging. What they share is that all are software-led or imaging-assisted device types that had been automatically placed in Class III by statute and have now been reclassified to Class II with special controls.

1. Over-the-Counter Device to Assess Risk of Sleep Apnea (21 CFR 868.2378)

FDA received a De Novo request from Samsung Electronics Co., Ltd. on 31 May 2023 for its Sleep Apnea Feature and issued the classification order on 6 February 2024; the device type is now codified at 868.2378.

The regulation defines the intended use very narrowly: for users who have never been diagnosed with sleep apnea, using a software algorithm to analyze sensor signals and provide a risk indication. The regulation also lists four prohibitions: it is not for standalone diagnosis, it does not replace conventional diagnostic methods such as polysomnography (PSG), it does not assist clinicians in diagnosing sleep disorders, and it is not to be used as an apnea monitor.

The special controls require clinical performance studies covering both symptomatic and asymptomatic populations, subgroup assessments by age, sex, body mass index and skin tone, and an accuracy comparison against a clinical comparator such as polysomnography. Machine learning algorithms must be clinically validated on data independent of the training set. Human factors testing must demonstrate that users can correctly use the device using the labeling alone, correctly understand the results and know when to seek medical care.

2. Software-Aided Adjunctive Diagnostic Device for Lesions Suspicious for Skin Cancer (21 CFR 878.1830)

This corresponds to the De Novo submitted by DermaSensor Inc. on 2 February 2022, classified in 2023, with the classification order issued on 12 January 2024; the device type is now codified at 878.1830.

The positioning of this regulation deserves separate attention. It specifies that the user is a physician who has not been trained in the clinical diagnosis of skin cancer, and that the purpose is adjunctive second-read interpretation after a suspicious lesion has been identified. It is not a standalone diagnostic tool and may not be used to confirm a clinical diagnosis.

One special control sets a hard threshold: the accuracy of interpretation with device assistance must be demonstrated to be superior to interpretation without the device. It further requires sensitivity of at least 90% for lesions with high metastatic potential, failing which a clinical justification must be provided. Testing must cover different skin types, risk factors and anatomical sites, with subgroup analyses. The order also allows FDA to require post-market surveillance data as appropriate.

3. Radiology Software for Referral of Findings Related to Fibrotic Lung Disease (21 CFR 892.2085)

This corresponds to the De Novo request submitted by Imvaria, Inc. on 29 June 2022 for the product Fibresolve; the order was issued on 12 January 2024 and the device type is now codified at 892.2085.

The regulation defines it as a prescription imaging device: it analyzes CT images and flags the presence of disease or a specific imaging sign, with output serving only as adjunctive information within a referral pathway and as part of the overall diagnostic assessment.

The risk table records both directions: a false positive could lead to incorrect or delayed management of a patient who in fact has another fibrotic lung disease, while a false negative could miss fibrotic lung disease. The core of the special controls is that the reference standard itself must be interpretable: the validation report must state the basis, method and appropriateness of the reference standard, not merely report an accuracy figure.

4. Spinal Imaging System for Neuraxial Procedures (21 CFR 868.1985)

This corresponds to the De Novo submitted by IntuiTap Medical, Inc. on 1 February 2022 for the VerTouch Spinal Imaging Device; the order was issued on 19 January 2024 and the device type is now codified at 868.1985.

It is a sensor-plus-software system intended to assist in identifying intervertebral space location for neuraxial procedures such as spinal anesthesia and lumbar puncture. The risks are stated concretely: inaccurate localization may cause pain, bleeding, unintentional dural puncture and injury to deeper structures, and poor use may also prolong procedure time.

Clinical performance testing must evaluate procedure success rate, procedure duration and adverse events (including pain, bleeding, unintentional dural puncture and injury to other spinal structures), compared against palpation or other localization methods. Labeling must state that the risks inherent in neuraxial procedures remain even when the device is used.

What the Orders Change and What They Do Not

Two layers need to be distinguished. The first is the classification decision: all four device types were already classified as Class II by FDA through De Novo in 2024, with classification effective dates of 12 January (two devices), 19 January and 6 February. The second is codification: these orders write those decisions into 21 CFR, and the orders take effect on the Federal Register publication date.

For industry, the real added value is citability. Previously, pathway assessment for similar products depended on the De Novo decision letters; now there are regulation numbers, codified text and a complete list of special controls that can be written directly into a regulatory strategy. The regulations also put on paper that subsequent similar devices may use the precedent device as a 510(k) predicate: the order text explicitly states that, following De Novo classification, the device may serve as a comparison basis for future similar devices.

What has not changed must also be stated clearly. The orders do not alter the premarket notification exemption status of these device types; whether an exemption applies is still determined by the FDA product classification database and the relevant exemption regulations. Nor do the orders retroactively change the 2024 classification effective dates. Products already cleared are unaffected.

AIMEILI Regulatory Interpretation

These four orders should be read as a coordinated tightening of the evidence standard for software-led and imaging-assisted devices rather than as a routine administrative cleanup. Three features run through all four regulations.

First, indications for use are drafted defensively. Each regulation pairs a narrow intended use with explicit prohibitions: the sleep apnea device may not claim to replace polysomnography or serve as an apnea monitor; the skin cancer software is positioned for physicians untrained in dermatologic oncology assessment and may not confirm a diagnosis; the fibrotic lung software output is adjunctive information within a referral pathway, not a diagnosis. For companies drafting 510(k) indications, the practical implication is that any claim beyond the codified wording will be treated as a new intended use and will attract additional evidence requirements.

Second, the special controls function as a de facto review checklist. Subgroup coverage by age, sex, BMI, skin tone, skin type and anatomical site; independent validation datasets for machine learning; a comparator arm; a superiority margin in the skin cancer regulation; a 90% sensitivity threshold for high-metastatic-potential lesions; and an explicit justification of the reference standard in the lung imaging regulation. In a 510(k), the submitter must demonstrate conformance with every applicable special control, so these elements are not optional enhancements.

Third, labeling and human factors are elevated to substantive requirements. The sleep apnea regulation writes the prohibition on replacing polysomnography into the labeling provisions, and the skin cancer regulation writes the superiority-over-unaided-interpretation requirement into the performance threshold. Human factors validation that demonstrates correct use and correct interpretation of results from the labeling alone is now a core evidentiary item for consumer-facing and physician-facing software alike.

Conclusion: the shift from De Novo to 510(k) lowers the procedural barrier but does not lower the scientific bar. The evidence package is essentially transferred from the De Novo decision into the special controls, and it must be met in full at the 510(k) stage.

Business Impact

For developers of similar devices in the pipeline, the predicate strategy is now clearer and cheaper: instead of a De Novo request, a 510(k) with a cited predicate becomes feasible, which typically shortens the interaction cycle and reduces the regulatory burden compared with a first-of-kind submission. However, the evidence plan must be built around the codified special controls from the protocol design stage, not retrofitted before submission.

For regulatory affairs and registration teams, the four regulation numbers should be added to the applicable standards and guidance matrix for any overlapping product line, and the labeling should be reviewed against the codified prohibitions and limitations of use. Claims that exceed the codified intended use should be removed or supported by separate evidence.

For CROs and clinical operations teams, study designs for these categories now have a fixed template: prespecified subgroups, an independent test set held separate from training data, an accepted clinical comparator (polysomnography, unaided interpretation, palpation), and a documented and justified reference standard.

Three Practical Actions

• If you have a similar product in development, transcribe the regulation numbers and the full special controls list into your evidence plan, and build subgroup, comparator and independent validation set requirements into the clinical protocol from the outset.
• If you develop software as a medical device, treat human factors testing and labeling wording as substantive requirements rather than documentation formalities, and verify that every prohibited claim in the regulation is excluded from your labeling and promotional material.
• If you develop imaging products, prepare the reference standard definition package in advance: both the fibrotic lung and spinal imaging regulations treat what constitutes ground truth and why it is appropriate as a review focal point rather than as supporting background.

Sources and Scope of Verification

Sources: four final orders of the U.S. Food and Drug Administration, all published in the Federal Register on 30 October 2026, docketed 29 October 2026.

• Federal Register — Medical Devices; Anesthesiology Devices; Classification of the Over-the-Counter Device to Assess Risk of Sleep Apnea (21 CFR 868.2378): https://www.federalregister.gov/d/2026-22182 (30 October 2026)
• Federal Register — Medical Devices; General and Plastic Surgery Devices; Classification of the Software-Aided Adjunctive Diagnostic Device for Use by Physicians on Lesions Suspicious for Skin Cancer (21 CFR 878.1830): https://www.federalregister.gov/d/2026-22179 (30 October 2026)
• Federal Register — Medical Devices; Radiology Devices; Classification of the Radiology Software for Referral of Findings Related to Fibrotic Lung Disease (21 CFR 892.2085): https://www.federalregister.gov/d/2026-22180 (30 October 2026)
• Federal Register — Medical Devices; Anesthesiology Devices; Classification of the Spinal Imaging System for Neuraxial Procedures (21 CFR 868.1985): https://www.federalregister.gov/d/2026-22181 (30 October 2026)
• U.S. Food and Drug Administration — CDRH New: News and Updates: https://www.fda.gov/medical-devices/medical-devices-news-and-events/cdrh-new-news-and-updates

Verification scope and access date: the FDA CDRH New updates page was accessed on 30 October 2026 to confirm the filing dates of the four orders; the full text of each order (public inspection PDF) was downloaded from federalregister.gov on the same day and checked item by item, including regulation numbers, intended use descriptions, classification effective dates, special control provisions and docket information. All dates, numbers and provisions cited in this article are taken from those original texts.

Scope note: this article is an interpretation of publicly available regulatory information and does not constitute a conclusion or commitment regarding any specific product, company or marketing application. Whether a device is exempt from premarket notification and whether additional materials are required are determined by the FDA product classification database and the official texts; information may change with subsequent official updates.