Non-Invasive Intracranial Pressure Monitoring: Where PPG Fits and Where It Doesn’t

PPG intracranial pressure monitoring is a promising research direction, but it is not a clinical replacement for invasive ICP monitoring today. Optical pulse waveforms can plausibly reflect changes in intracranial compliance and vascular pulsatility, yet the evidence base is still dominated by modeling, bench testing, and early observational work rather than large prospective validation studies.

What clinicians mean by non-invasive ICP monitoring

In neurocritical care, the reference standard for ICP is still invasive measurement through an external ventricular drain or an intraparenchymal sensor. That matters because clinicians do not only need a trend line. They often need an accurate pressure value, a reliable alarm for dangerous elevation, and sometimes a route for cerebrospinal fluid drainage.

That is why most non-invasive tools are better described as screening, triage, or adjunctive monitoring methods. They may help estimate risk of elevated ICP, add context to bedside assessment, or support longitudinal research. They do not yet replace invasive devices when treatment decisions depend on precise ICP data. Recent reviews continue to make that point clearly, including Robba et al. in Acta Neurologica Scandinavica and a 2023 review in Journal of Clinical Medicine.[1][2]

If you are building an evidence stack around optical biosignals, it helps to separate three questions:

1. Can a PPG signal change when ICP changes?
2. Can an algorithm estimate ICP or detect elevated ICP from those changes?
3. Can that estimate generalize across settings, patient groups, and hardware?

PPG looks most credible on the first question, somewhat promising on the second, and still unproven on the third.

Why PPG is mechanistically plausible

Photoplethysmography measures pulsatile optical changes caused by blood volume variation in tissue. In plain language, it is a light-based way to capture how the pulse wave moves through the vascular bed. Because intracranial pressure is coupled to cerebral blood volume, cerebrospinal fluid dynamics, venous outflow, and cranial compliance, it is reasonable to ask whether ICP-related physiology leaves a signature in an optical waveform.

The mechanism is plausible for several reasons:

• Rising ICP can alter intracranial compliance and the transmission of arterial pulsations.
• Changes in venous drainage or cerebrovascular tone can reshape the local pulse waveform.
• Optical signals are sensitive to amplitude, timing, and baseline shifts that may track these hemodynamic changes.

A 2023 computational study by Liu et al. supports that plausibility. In their model, worsening intracranial compliance and higher ICP changed value-related PPG features such as maximum, minimum, mean, and amplitude, while some shape-related features changed far less. The same study also found that age and measurement territory mattered, which is a useful warning: even if the signal contains ICP information, it is not isolated from anatomy or physiology.[3]

That last point is the biggest reality check. PPG is not an ICP sensor in the same way a ventricular catheter is an ICP sensor. It is an indirect biosignal. The waveform is also influenced by blood pressure, heart rate, carbon dioxide, vasopressor use, temperature, motion, sensor contact, skin and soft-tissue properties, vascular stiffness, and site selection. Any algorithm that ignores those confounders will look better in a small dataset than it will in real clinical use.

For a broader primer on optical signal behavior, the ChatPPG learn hub and our algorithm library are better starting points than any single headline study.

What the evidence says right now

The current literature on non-invasive ICP monitoring is broad, but few approaches have matured into dependable routine care tools. Reviews consistently conclude that invasive monitoring remains the reference method and that no non-invasive modality has yet shown enough accuracy, reproducibility, and clinical validation to replace it.[1][2]

Within the PPG-specific literature, the most encouraging studies are also the ones that require the most caution.

1. Bench and computational work is encouraging, but it is not bedside validation

Abay et al. evaluated a non-invasive PPG-based ICP sensor in an in vitro cerebral hemodynamics model. Their system showed high correlation with the reference signal and relatively low root mean square error under controlled conditions.[4] That is useful because it suggests the optical features can carry pressure-related information when noise, movement, and biological variability are constrained.

But in vitro performance is not the same as clinical performance. A bench model cannot fully reproduce skull properties, perfusion heterogeneity, edema, sedation changes, vasoactive drugs, or the messy signal loss seen in real patients.

Liu et al. add mechanistic support from a computational angle, again showing that simulated ICP changes can alter PPG waveform values.[3] Computational evidence is helpful for hypothesis building and feature selection. It is not proof that a wearable can measure bedside ICP accurately enough for treatment decisions.

2. The hardest part is generalization

A model can correlate with ICP in a single center and still fail when moved to another ward, sensor geometry, disease group, or skin type. This is especially important for PPG ICP work because the signal path is indirect. A model trained on traumatic brain injury patients in one setup may not transfer to hydrocephalus, idiopathic intracranial hypertension, post-neurosurgical patients, or outpatient screening.

The likely near-term value of PPG is not one universal number that works everywhere. It is narrower use cases, such as trend tracking in a controlled cohort, screening for sustained change, or adding a probabilistic signal to a multimodal model.

3. Correlation is not enough

A strong Pearson correlation is not the same as clinical agreement. If a system will be used to flag ICP above 20 mmHg, authors should report threshold performance, calibration, Bland-Altman limits of agreement, failure rates, and subgroup behavior. If it will be used to follow trends, authors should show whether direction and magnitude of change are captured during clinically relevant events.

That is where a lot of non-invasive ICP papers still come up short. The most interesting results often show signal association, not decision-grade reliability.

Where PPG fits today

If the goal is realistic product strategy or study design, PPG fits best in four places.

Research on physiology and signal features

PPG is well suited to exploratory work on how intracranial dynamics may influence peripheral or extracranial optical signals. That includes feature engineering, artifact handling, multimodal fusion, and disease-specific modeling.

Multimodal risk estimation

PPG makes more sense as one input among several. In practice, a realistic stack might combine optical features with blood pressure, demographic factors, neurologic exam findings, imaging context, or other bedside measures. That is more defensible than claiming a single optical channel can solve ICP monitoring alone. Our conditions hub and charts library are useful for framing disease context and thresholds.

Longitudinal or lower-acuity monitoring research

Wearable or semi-wearable optical sensing may be valuable where invasive monitors are inappropriate, unavailable, or too risky. That could include observational follow-up, symptom-linked trend studies, or home-to-clinic monitoring concepts. The keyword there is research. It does not yet mean diagnostic clearance or treatment guidance. For device constraints, see our wearables hub.

Early warning or triage support

A future PPG system may prove helpful as an alerting layer that says, "this patient looks different from baseline" or "this pattern deserves confirmatory testing." That is a very different claim from saying it measures true ICP in mmHg with bedside equivalence.

Where PPG does not fit today

PPG should not currently be marketed or interpreted as:

• a replacement for an external ventricular drain or intraparenchymal monitor in critical care
• a stand-alone diagnostic test for intracranial hypertension
• a reliable outpatient wearable that confirms elevated ICP from a consumer-style reading
• a single model that generalizes across ages, pathologies, and sensor placements without recalibration
• a safe basis for decisions such as CSF drainage, shunt adjustment, or decompressive intervention without confirmatory clinical evidence

That is not a knock on the signal. It is a reminder that the bar is high. ICP monitoring is clinically important precisely because the consequences of error are high.

How to design a credible PPG ICP study

If you want to advance the field without overpromising, a strong study design matters more than a flashy correlation coefficient.

Use the right reference

When ethically justified, compare against invasive ICP, not a softer surrogate. If the clinical question is elevated ICP detection, define the reference window and label timing carefully.

Predefine the clinical use case

Choose one primary task:

• absolute ICP estimation
• threshold detection
• trend detection
• event prediction

Do not blur them after the results are in. A model that tracks direction may still be poor at absolute mmHg estimation.

Split by patient, not just by beat

PPG datasets are dense. If training and test sets share data from the same patient, performance can look inflated. External validation across sites and hardware is much more convincing than random beat-level splits.

Report agreement, calibration, and failure

Publish Bland-Altman limits, calibration plots, subgroup analyses, and the proportion of unusable windows. In wearable sensing, failure rate is part of performance.

Stress the confounders

Test body position, age, blood pressure swings, respiratory changes, sedation, movement artifact, vasopressors, anemia, skin tone, edema, and site-specific sensor placement. A robust model has to survive the variables that dominate real care.

Keep the claim narrow

A narrow true claim beats a broad false one. "This model identified sustained upward ICP trends in a monitored TBI cohort" is better science than "PPG non-invasively measures ICP."

Bottom line

PPG intracranial pressure monitoring is scientifically plausible and increasingly interesting, especially as signal processing and wearable optics improve. But the honest state of the art is that PPG is best viewed as an indirect, research-stage component of non-invasive ICP estimation, not as a replacement for invasive monitoring or a standalone clinical diagnostic today.

If you want the short version, use this: PPG may help detect pressure-related physiology, but it has not yet earned the right to claim bedside ICP equivalence. For more clinical context and adjacent topics, browse the ChatPPG blog.

FAQs

Can PPG directly measure intracranial pressure?

No. PPG captures optical pulse-related changes in tissue blood volume. Any ICP estimate derived from it is indirect and model-based.

Is there evidence that PPG waveforms change when ICP changes?

Yes, there is early evidence from computational and bench studies that ICP-related physiology can alter PPG features. That supports plausibility, but it is not the same as validated clinical accuracy.[3][4]

Could a wearable eventually help screen for elevated ICP?

Possibly, but the most realistic near-term role is screening or trend detection in a defined context, followed by confirmatory testing. It is not ready to replace invasive monitoring.

Why is generalization so hard for PPG ICP models?

Because PPG is influenced by many variables besides ICP, including blood pressure, age, sensor site, motion, vascular tone, and tissue properties. Models that look strong in one dataset may not transfer cleanly to another.

What outcome should future studies prioritize?

They should prioritize a clearly defined clinical task, such as detecting sustained ICP above a threshold or tracking change over time, then validate it prospectively against invasive reference measurements.

What would change the field meaningfully?

Large prospective multicenter studies, external validation across hardware types, transparent reporting of agreement and failure rates, and evidence that the signal improves real clinical decisions.

References

1. Robba C, Bacigaluppi S, Cardim D, Donnelly J, Bertuccio A, Czosnyka M. Non-Invasive Assessment of Intracranial Pressure. Acta Neurol Scand. 2016;134(1):4-21. DOI: https://doi.org/10.1111/ane.12527
2. Non-Invasive Intracranial Pressure Monitoring. J Clin Med. 2023;12(6):2209. DOI: https://doi.org/10.3390/jcm12062209
3. Liu H, Pan F, Lei X, Hui J, Gong X, et al. Effect of intracranial pressure on photoplethysmographic waveform in different cerebral perfusion territories: A computational study. Front Physiol. 2023;14:1085871. DOI: https://doi.org/10.3389/fphys.2023.1085871
4. Abay TY, Phillips JP, Uff C, Roldan M, Kyriacou PA. In Vitro Evaluation of a Non-Invasive Photoplethysmography Based Intracranial Pressure Sensor. Appl Sci. 2023;13(1):534. DOI: https://doi.org/10.3390/app13010534