FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Apex Insurance Brokers
Speak to a brokerGet a quote →

Aipedia: Occupational health providers

AI in health surveillance: audiometry, spirometry and skin checks

AI can analyse audiograms, spirometry and skin images in health surveillance, but a competent clinician should review flagged and borderline results. A missed early sign of noise-induced hearing loss or occupational asthma can lead to claims from workers and employers.

In short

Employers must provide suitable health surveillance where risk assessments show a need, including hearing tests under the Control of Noise at Work Regulations 2005 and surveillance under COSHH. OH providers use automated software to categorise audiograms, interpret spirometry and screen skin images. Errors can leave a worker exposed to further harm and an employer without the information it needed to act. Health surveillance decisions with significant effects on workers fall within UK GDPR Articles 22A to 22D if made solely by automated means. A clinician should sign off results and refer abnormal findings.

How AI is used in surveillance

Statutory health surveillance generates large volumes of standardised tests. Audiometry software categorises audiograms against reference levels and flags shifts from baseline. Spirometry software interprets flow volume curves and compares them with predicted values. Some providers use image analysis to screen photographs of hands for dermatitis. Mobile surveillance units may run tests with technicians, sometimes with automated results sent to employers and a clinician reviewing only flagged cases.

The legal duty to provide surveillance sits with the employer, but the employer may rely heavily on the OH provider to design the programme, interpret the results and tell it when action is needed. That reliance is what turns an automated misreading into a professional liability question for the provider. It also means the provider's contract should say clearly which results a clinician reviews and how quickly abnormal findings are reported.

The duties that apply

What can go wrong

Illustrative scenario: a mobile unit runs annual audiometry for a manufacturer. The software rates a worker's audiogram as acceptable because it compares the result with population references rather than the worker's own baseline, and no clinician reviews it. Over several years the worker's hearing deteriorates. When the worker brings a claim against the employer, the employer asks why the OH provider's surveillance did not flag the change, and looks to the provider for a contribution.

Similar risks include spirometry interpreted against inappropriate reference values, poor-quality tests accepted as valid, missed patterns of work-related symptoms in questionnaires, and skin images taken in poor light. Surveillance claims can arise many years after the test, so records and the logic used at the time matter.

How PI usually responds

A claim that surveillance was carried out or interpreted negligently is usually a professional indemnity matter for the OH provider, subject to its terms. Because the worker's injury may be the basis of the claim, check how the policy treats bodily injury and contribution claims from employers. Long-tail exposure means the policy in force when the claim is made matters, and run-off cover matters if the provider closes. Policies may also assume that a clinician reviewed the results. See does OH indemnity cover AI mistakes, claims-made versus occurrence and the HSE.

Practical controls

Related: AI in safety-critical medicals, automated decisions under DUAA and the occupational health AI hub.

How Apex helps

Apex arranges PI for occupational health providers that carry out health surveillance. A broker reviews how your policy treats automated test interpretation, bodily injury and long-tail claims; see OH professional indemnity.

Sources

Frequently asked

Can audiometry software decide results without a clinician?

Software can categorise results, but the noise regulations require a doctor to examine a worker where hearing damage is found, and a clinician should review flagged and borderline results. Solely automated significant decisions on health data also engage UK GDPR Articles 22A to 22D.

Why compare with the worker's baseline?

Noise-induced hearing loss can show as a change over time. A result that looks acceptable against population references may still be a significant shift for that worker. Baseline comparison and clinician review reduce the chance of missing early damage.

How long can surveillance claims take to appear?

Many years. COSHH requires health records to be kept for at least 40 years, reflecting how long occupational disease can take to develop. OH providers should think about long-tail claims and run-off cover when they close or change policy.

Who holds the legal duty for health surveillance?

The employer. The noise and COSHH regulations place the surveillance duty on the employer, which may rely on an OH provider to run it. If the provider's interpretation is flawed, the employer may look to the provider for a contribution, so contracts should set out who reviews which results.

Check your indemnity covers how you use AI

Tell us about your practice and the work you do. A broker will review how your current cover treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start an indemnity quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.