Table 2 Barriers to target identification referenced across studies

Data acquisition

TCD

Anomalous/contaminated data (e.g., movement, electrocautery, physiologically impossible velocity)

7

Manual removable of artifacts, utilize only unilateral recordings, or patient exclusion

In cardiac surgery, this affects pre/post-CPB periods more due to high electrocautery activity. Patient attrition and signal loss related to this cause were up to 23% [51] and 36% respectively [60]. In brain injury, attrition rate related to this cause was 7% [33]. Manual removal might not perfectly rule out contaminated signals. [52, 56, 57, 62]

Lack of transcranial window

4

Patient exclusion

Minimally impacts recruitment rate (4–7% of recruited patients) and could be identified early in the research

[33, 61,62,63]

Forehead obstruction (e.g., monitors, dressing, hemicraniectomy)

1

Patient exclusion or use only unilateral recordings

Specific prevalence was not reported [33]

NIRS

Contaminated signals (metabolic/physiological changes incompatible with the recording mechanism)

2

Patient exclusion

Patient attrition related to this cause was few (6%)[53]; could be inherent to the technology thereby impact is masked, specific populations might be more affected and should be identified [44]

5/6/2025 7:19:00 PM

Sensor adhesion issues

1

Remove recordings

Specific prevalence was not reported [39]

Optode malfunction

1

Patient exclusion or use only unilateral recordings

Affected 1% of patients [38]

Recording failure

1

Specific prevalence was not reported [37]

Equipment unavailability

1

Specific prevalence was not reported [40]

MAP

Anomalous or contaminated data

3

Manual or auto/algorithmic MAP cleaning

Affected up to 15% of participants [40, 41, 43]

Software failure (loss of continuous MAP data)

1

Patient exclusion and data cleaning

Affected up to 23% participants [62]

Other or unspecified

Equipment unavailability, no high-fidelity recording, data loss while transferring, or protocol miscommunication

4

Patient exclusion, re-training and equipping sites, extract data immediately after recording, review protocol and improve documentation

Affected 3% to 53% of eligible participants depending on when the protocol is reviewed and revised [37, 40, 43, 48]

Anomalous or contaminated data

3

Data cleaning or patient exclusion

Affected 14% of data recorded [37] and 23–26% of participants [55, 58]

Damaged recordings or other recording complications

2

Patient exclusion

Affected 9% [45] and 4% of all eligible participants [37]

ICU discharge before study completion

1

Modify study design/timing

Specific prevalence was not reported [48]

Target calculation

No observable MAP limits—MAP always above/outside acceptable thresholds

15

Define LLA as the lowest COx above the threshold, adjusting threshold values

Affected 2–100% participants depending on study methods. Gergelé et al.[52] tried calculating LLA with 15-min long recording and was unsuccessful. Møller et al.[35] tried calculating LLA in patients with bacterial meningitis in the first 24 h after diagnostic lumbar puncture and was also unsuccessful until later in their recovery. Asides from these two studies, loss related to this cause is lower than 35%. Yield appears lower for ULA vs LLA [29]. [33, 43,44,45, 49, 51, 53, 54, 63, 65,66,67]

No observable MAP limits—MAP always below/within acceptable thresholds

7

No steps needed as assumed part of healthy variation; prevalence should be transparently reported

Affected 2–60% of participants, regression studies more heavily affected; threshold values influenced detection rates; surgical studies generally had better identification rates than ICU studies (2–18% loss versus 26–27% [correlation approach only]) [29, 43, 49, 53, 54, 63, 67]

Short recording duration

5

Attempts made to calculate but if unsuccessful patients will be excluded

Affected 10% of participants [33, 34, 38, 44, 54]

Fluctuating thresholds

1

N/A

Affected 15% (Mx2s) and 16% (Mx10s) of participants [51]