Psychiatric inpatient wards have a monitoring problem that most hospitals do not. Patients in acute mental health units require frequent observation, sometimes every 15 minutes, sometimes continuously. Staff check through bedroom door windows to confirm the patient is breathing, awake, or not harming themselves. These checks happen around the clock. They disrupt sleep. They can agitate patients who are already in crisis. And when a ward is short-staffed, they get delayed.
The question driving a growing body of research: can camera-based systems fill the gaps between physical checks without making things worse?
"Oxehealth technology was considered acceptable to both staff and patients if the technology was used to detect deterioration and improve patient's safety providing patient's privacy was not invaded." — Boughen et al., BMC Psychiatry (2023)
Why psychiatric wards are different from general hospital monitoring
In a general medical ward, vital sign monitoring is straightforward. Patients wear pulse oximeters, blood pressure cuffs cycle automatically, and continuous telemetry picks up cardiac abnormalities. Psychiatric inpatient settings operate under different constraints. Many patients are detained under mental health legislation. Some are at risk of self-harm or suicide. Others experience psychosis, paranoia, or severe agitation. Attaching wearable devices to these patients is often impractical or clinically contraindicated. A pulse oximeter can become a ligature risk. Wired monitoring restricts movement and can escalate distress.
This creates a gap. General wards have continuous physiological data. Psychiatric wards have periodic human observation, supplemented by whatever the patient is willing to tolerate. That gap has consequences. A 2022 review by Bowers et al. documented that most serious incidents on psychiatric wards, including self-harm and sudden medical deterioration, happen between observation intervals or during nighttime when staff presence is reduced.
Vision-based patient monitoring attempts to bridge this gap by placing infrared-sensitive cameras in patient rooms. The camera captures movement and, in some implementations, extracts pulse and respiratory rate from subtle skin color changes using principles similar to remote photoplethysmography. Staff view the feed from a nursing station, and the system generates alerts for concerning patterns like prolonged stillness or a patient leaving their bed at an unusual hour.
How these systems work in practice
The most extensively deployed system in psychiatric settings is Oxevision, developed by Oxehealth and implemented across multiple National Health Service mental health trusts in the United Kingdom. The system uses an infrared-sensitive camera mounted in the patient's bedroom. It provides three main functions: remote visual observation (a brief encrypted video check of the room), contactless vital signs measurement (pulse and respiratory rate extracted from the video feed), and automated behavioral alerts.
The camera captures 15-second encrypted video clips that staff can view on demand. Video is stored for up to 24 hours before being overwritten. Staff can check whether a patient is in bed, moving normally, or in distress without physically walking to the door and looking through the window. The system also flags events like a patient spending an unusually long time in the bathroom, which could indicate a medical emergency or self-harm attempt.
According to the Agency for Care Effectiveness in Singapore, which conducted an independent horizon scan of the technology in 2024, Oxevision is CE-marked as a medical device and has been through data protection assessments against GDPR, the UK Data Protection Act 2018, and Article 8 of the European Convention on Human Rights.
| Monitoring approach | Contact required | Continuous data | Sleep disruption | Ligature risk | Staff time per check | Current psychiatric use |
|---|---|---|---|---|---|---|
| Physical door checks (every 15 min) | None (visual through window) | No | Yes, significant | None | 2-5 minutes per round | Standard of care everywhere |
| Wearable pulse oximeter | Finger clip | Yes | Moderate (device discomfort) | Potential (cord/band) | Minimal once placed | Limited, general medical only |
| Wearable biosensor patch | Adhesive | Yes | Low | Low | Minimal once placed | Rare in psychiatric settings |
| Vision-based monitoring (Oxevision) | None | Semi-continuous (on demand) | Reduced (no door opening) | None | Seconds per remote check | Deployed in 15+ NHS trusts |
| rPPG via smartphone/tablet | None (requires facing camera) | No (spot check) | Low | None | 30-60 seconds per reading | Research phase only |
| Bed sensor mat | Under mattress | Yes | None | None | Minimal | Emerging, limited psychiatric data |
What the evidence shows so far
The published research on vision-based monitoring in psychiatric settings is growing but still limited in quality. A systematic review published in BMC Medicine in December 2024 by Sheridan Rains et al. examined 32 studies on surveillance technologies in inpatient mental health settings. The review covered CCTV, vision-based monitoring systems, body-worn cameras, GPS tracking, and wearable sensors. Their findings were sobering: 50% of included studies were rated low quality, and 28.1% declared conflicts of interest, typically involving researchers affiliated with or funded by the technology companies being evaluated.
The review found that quantitative evidence regarding the impact of these technologies on self-harm, violence, aggression, and care quality was "inconsistent or weak." One notable finding: a widely cited paper by Kekic and Rose reporting that Oxevision reduced self-harm across five NHS trusts was retracted in 2024 due to undeclared conflicts of interest. The retraction removed one of the strongest quantitative claims supporting the technology.
That does not mean the technology has no value. Qualitative evidence tells a more nuanced story. Boughen et al. published a qualitative service evaluation in BMC Psychiatry in 2023, based on interviews with 12 staff members and 12 patients at Broadmoor Hospital, a high-secure forensic psychiatric facility. Both groups found the technology generally acceptable when used to detect deterioration and improve safety. Staff valued the ability to confirm a patient's status without opening a door, particularly during nighttime hours. Patients appreciated fewer physical disruptions to their sleep.
But acceptance was not universal. Patients described "big brother" concerns. Several could not understand why the hospital needed both camera monitoring and physical checks. Staff were clear that the technology should supplement, not replace, in-person observation. The therapeutic relationship built through face-to-face contact was something neither group wanted to lose.
A separate study published in the Journal of Mental Health Nursing by Butler et al. (2023) examined Oxevision implementation in an acute adult mental health ward. Their qualitative findings echoed Broadmoor: the system helped staff detect incidents more quickly, particularly overnight when staffing levels drop, but raised consistent questions about privacy, consent, and whether detained patients can meaningfully consent to being monitored by camera.
The consent problem
This is where psychiatric monitoring diverges from other clinical applications. In a general hospital, a patient can refuse a blood pressure check. In many psychiatric inpatient settings, patients are detained under mental health legislation and may not have the option to decline monitoring. Sheridan Rains et al. flagged this as a significant gap in the literature. Most studies did not adequately address whether patients provided informed consent to being monitored, how consent was obtained from patients who were acutely unwell, or what happened when a patient objected.
The Stop Oxevision campaign, led by patient advocacy groups in the UK, has raised several concerns: that consent processes are inadequate for acutely unwell patients, that camera monitoring could worsen paranoia and distress, and that cost-saving motivations could lead to staffing reductions disguised as technological improvement.
Where contactless vital signs add clinical value
Beyond safety monitoring, there is a clinical rationale for continuous vital sign data in psychiatric populations that is less controversial than the surveillance question.
Psychiatric inpatients have high rates of undetected physical health problems. A 2020 meta-analysis by Launders et al. in the British Journal of Psychiatry found that people with serious mental illness have a life expectancy 15 to 20 years shorter than the general population, driven largely by undiagnosed and undertreated cardiovascular and metabolic conditions. On psychiatric wards, vital signs are often checked infrequently because the clinical focus is on mental state rather than physical health.
Antipsychotic medications carry known cardiac risks, including QT prolongation and metabolic syndrome. Lithium requires careful monitoring of hydration status. Physical restraint and rapid tranquilization carry acute cardiovascular risks. Having continuous, passive heart rate and respiratory rate data could flag early signs of medication side effects, autonomic instability, or physical deterioration in patients who may not report symptoms because of their psychiatric condition.
This is where rPPG-based approaches have a potential role that extends beyond what current camera monitoring systems offer. Existing vision-based systems like Oxevision can detect movement and measure basic pulse and respiratory rate. More advanced rPPG algorithms can potentially extract heart rate variability, respiratory patterns, and relative changes in blood oxygen saturation from standard video. For a population where wearable monitoring is often impractical, that additional physiological data could be clinically meaningful.
The staffing question nobody wants to answer directly
The elephant in the room with psychiatric monitoring technology is cost. NHS mental health trusts face chronic staffing shortages. One-to-one observation of high-risk patients is extremely expensive, requiring a dedicated staff member for each patient around the clock. If a camera can reduce the need for continuous observation, the financial savings are substantial.
Research has acknowledged this driver. An economic evaluation referenced in the ACE Singapore horizon scan estimated cost savings from reduced observation hours after Oxevision implementation. But patient advocacy groups and some clinicians argue that framing technology as a replacement for human staff is dangerous in a setting where therapeutic relationships are the primary treatment modality.
The honest answer, based on current evidence, is that nobody has rigorously demonstrated either outcome. We do not yet have high-quality randomized trials showing that vision-based monitoring is safe as a partial substitute for physical observation, nor do we have strong evidence that it cannot be. The Sheridan Rains systematic review concluded that "further independent coproduced research is needed to more comprehensively evaluate the impact of surveillance technologies in inpatient settings."
What comes next for psychiatric contactless monitoring
The next phase of this technology will likely be shaped by two forces. First, regulatory scrutiny. The UK's Care Quality Commission and similar bodies in other countries are beginning to develop frameworks for surveillance technology in mental health settings. These frameworks will determine how and whether these systems can be deployed.
Second, algorithmic improvement. Current systems provide relatively basic physiological data. As rPPG algorithms improve in accuracy and robustness under real-world conditions including low light, patient movement, and diverse skin tones, the clinical value proposition shifts from "can we see if the patient is moving" to "can we detect physiological deterioration before it becomes an emergency." That is a harder problem to solve, but a more defensible clinical justification.
Circadify has developed camera-based vital sign measurement technology that could be applied in behavioral health settings. The ability to passively extract heart rate, respiratory rate, and other physiological signals from video has applications in environments where traditional wearable monitoring is difficult to implement. The technical capability exists. The clinical validation and ethical framework around its use in psychiatric populations is what needs to catch up.
Frequently asked questions
How does vision-based monitoring work in psychiatric inpatient settings?
Vision-based systems use infrared-sensitive cameras installed in patient rooms to detect movement, measure pulse and breathing rates from subtle skin changes, and alert staff to potentially dangerous situations like prolonged inactivity or unusual behavior. All of this happens without physical contact with the patient.
Do patients in psychiatric wards accept camera-based monitoring?
Research is mixed. A 2023 qualitative study at Broadmoor Hospital by Boughen et al. found most staff and patients considered passive monitoring acceptable when used for safety purposes, but acceptance dropped when patients did not fully understand the technology. Privacy concerns remain a consistent theme.
Can contactless monitoring replace physical observation checks in mental health wards?
Current evidence does not support full replacement. Staff in multiple evaluations have said the technology should supplement rather than replace in-person checks. Physical observations allow clinical judgment and therapeutic interaction that cameras cannot replicate.
What are the ethical concerns around camera monitoring in psychiatric settings?
The main concerns include patient consent (particularly for detained patients), privacy invasion, potential for worsening paranoia or distress, data security, and the risk that technology could justify staffing reductions rather than genuinely improve care.
Related articles
- rPPG Technology and Mental Health Screening — How remote photoplethysmography is being explored for mental health assessment and emotional state detection.
- Contactless Vital Signs for Elderly Aging in Place — Camera-based monitoring for elderly populations living independently at home.
- Privacy and Data Security in Camera-Based Health Monitoring — An analysis of the privacy and regulatory landscape surrounding camera-based vital sign technology.