# Passive sensing of gait and medication-related fluctuations in Parkinson's disease.

- **DOI:** [10.1186/s12984-026-02068-6](https://doi.org/10.1186/s12984-026-02068-6)
- **Rating:** Sound (3.2 out of 5)
- **Verdict:** Passive sensors detect selected Parkinsonian gait and medication-state differences in a living lab, but same-data calibration limits confidence in spatial agreement.
- **Framework version:** v5.9
- **Pipeline version:** v0.4
- **Evaluation date:** 2026-10-08
- **Evaluation page:** https://www.nabu.science/paper/10.1186%2Fs12984-026-02068-6

## Key claims

1. Stride time demonstrated the highest agreement between devices (r = 0.903), while stride length was weaker (r = 0.779).
   - Support: Supported
   - Type: Primary finding
   - Basis: inter-device statistics agree, but evidence comes from a short, single-site 4-m task
   - Location: Abstract; Results, Agreement Between Devices; Figure 3
2. both systems demonstrate sensitivity to gait patterns and medication-dependent changes, supporting their use for longitudinal, real-world monitoring of motor symptoms.
   - Support: Tentative
   - Type: Secondary
   - Basis: the camera did not separate medication states after correction, and the design was neither longitudinal nor home-based
   - Location: Abstract, Conclusions
3. Camera-derived step length asymmetry showed a moderate positive correlation with UPDRS asymmetry (r = 0.547, p = 0.002).
   - Support: Tentative
   - Type: Secondary
   - Basis: pooled repeated states and 72.6% relative error materially limit the single-session correlation
   - Location: Results, Symmetry Index Comparison; Figure 5
4. stride length from both Kinect and radar distinguished PD OFF from HC, and radar additionally differentiated ON from OFF.
   - Support: Supported
   - Type: Secondary
   - Basis: FDR-corrected group differences support direction, though the small living-lab sample limits inference
   - Location: Abstract, Results; Results, Group and State Differences; Figure 4a
5. radar offers enhanced sensitivity to dopaminergic state changes, particularly through stride length evaluation
   - Support: Supported
   - Type: Secondary
   - Basis: radar alone showed a corrected ON–OFF stride-length difference, but the finding is from 15 patients
   - Location: Conclusion; Results, Group and State Differences; Figure 4a

## Strengths

- **Paired medication-state sensing design** Participants with Parkinson’s disease were assessed in both ON and OFF states during their usual medication cycle, allowing within-patient analysis of natural fluctuations.
  - Dimension: methodological rigour
  - Location: Results, Study Procedures; Methods, Experimental Setup and Protocol
- **Concurrent multimodal agreement assessment** Camera, radar, and floor systems operated simultaneously, with ICC and Bland–Altman analyses supplementing correlations for inter-device agreement.
  - Dimension: methodological rigour
  - Location: Results, Study Procedures and Agreement Between Devices; Methods, Statistical Analysis
- **Limitations tied to future validation** The Discussion identifies the non-home setting, instructed 4-m walk, mild and predominantly male cohort, and need for wearable comparisons as constraints on interpretation and next steps.
  - Dimension: positioning
  - Location: Discussion, final paragraph

## Limitations

- **Calibration and evaluation share trials** Camera spatial calibration constants were derived from the study trials and applied before agreement analysis on those same data. This makes the near-zero calibrated bias and spatial agreement optimistic.
  - Dimension: methodological rigour
  - Location: Methods, Camera Data; Results, Agreement Between Devices
- **Real-world conclusion exceeds design** The Abstract supports longitudinal, real-world monitoring, but the study collected single-session instructed walks in a living laboratory rather than continuous home data.
  - Dimension: reporting
  - Location: Abstract, Conclusions; Discussion, limitations paragraph
- **Restricted cohort and walking task** The evidence comes from 15 mild-PD participants at one site using a 4-m straight path, with male predominance and no assessment of free-living walking or turning.
  - Dimension: application pathway
  - Location: Methods, Participants and Ethical Information; Results, Study Procedures; Discussion, limitations paragraph

## Reliability checks

Caveats (4 of 4 checks).

- **Results check:** Noted
  - stride length from both Kinect and radar distinguished PD OFF from HC, and radar additionally differentiated ON from OFF: "Nevertheless, stride length from both Kinect and radar distinguished PD OFF from HC (camera q = 0.020; radar q = 0.005), and radar additionally differentiated ON from OFF (q = 0.020)." (Abstract — Results) vs "Similarly, the camera showed a weaker but still significant difference between HC and PD OFF participants (q = 0.025)" (Results — Group and State Differences). The abstract reports camera HC–PD OFF q = 0.020 whereas the Results report q = 0.025; both values are below 0.05. → Conclusion holds
  - stride length from both Kinect and radar distinguished PD OFF from HC, and radar additionally differentiated ON from OFF: "Nevertheless, stride length from both Kinect and radar distinguished PD OFF from HC (camera q = 0.020; radar q = 0.005), and radar additionally differentiated ON from OFF (q = 0.020)." (Abstract — Results) vs "For stride length (Fig. 4a), the radar successfully distinguished differences between HC and PD OFF participants (q = 0.005) and between PD ON and OFF states (q = 0.024)." (Results — Group and State Differences). The abstract reports radar ON–OFF q = 0.020 whereas the Results report q = 0.024; both values are below 0.05. → Conclusion holds
  - both systems demonstrate sensitivity to gait patterns and medication-dependent changes, supporting their use for longitudinal, real-world monitoring of motor symptoms: "Although some spatial metrics show device discrepancies, both systems demonstrate sensitivity to gait patterns and medication-dependent changes, supporting their use for longitudinal, real-world monitoring of motor symptoms." (Abstract — Conclusions) vs "Similarly, the camera showed a weaker but still significant difference between HC and PD OFF participants (q = 0.025) but could not significantly discriminate between ON and OFF states after correction for multiple comparisons (p = 0.018, q = 0.051)." (Results — Group and State Differences). The claim attributes demonstrated medication-state sensitivity to both systems, but the Results explicitly state that the camera ON–OFF comparison does not survive correction (q = 0.051). → Not supported by the paper's own data
- **Reporting and transparency:** Noted. The abstract's ground-truth label does not convey the floor reference's restriction to torso speed, and the participant methods contain no ethics-approval or consent declaration.
  - Conflict of interest: not stated · Ethics approval: not stated · Preregistration: not stated · Data availability: not stated · Code availability: not stated
- **Reference integrity:** Clean
  - 62 references in manuscript
  - 61 of 61 checkable references found in an index
  - 1 are books, websites or datasets — counted, but not index-checkable
  - References not verified automatically are resolved manually within 24 hours.

## Post-publication record

- **Status:** Clean
- **Summary:** No retraction or expression of concern found in Crossref (includes Retraction Watch).
- **Last checked:** 2026-10-17
