AI Science Tracker
EN

Politecnico di Torino

3 discoveries from 2021 to 2026

Discoveries

A policy exploration reset helped a robot controller adapt in simulation

Source check: Source checked. Not independent scientific confirmation.

Finding: In 30 simulation seeds, restoring an exploration parameter helped pretrained PPO controllers adapt to a reaching task after a robot joint was locked. A physical deployment also favored the reset condition for the single simulation-selected policy tested per method.

Technology: Proximal Policy Optimization with Gaussian policies

Publication: Preprint · Test context: direct test. Catalog categories, not an evidence grade.

AI contribution
PPO-trained policies were tested with an intervention on their exploration distribution, showing a change in adaptation after the simulated task shifted.
Test and result
The simulation compared continued training with a one-time exploration reset and different entropy penalties. Each physical policy was tested against the same 50 target points.
Limitation / still unknown
The physical experiment used a single simulation-selected seed per method and consecutive target points. The result is specific to the studied Gaussian policy parameterization and adaptation task.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://arxiv.org/abs/2610.36309v1 · https://arxiv.org/html/2610.36309v1

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 21 Nov 2026. Review scope: §§3 and 4.3, Figure 6, Table 1 and hardware sampling protocol.

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 3/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Characterization of Infants' General Movements Using a Commercial RGB-Depth Sensor and a Deep Neural Network Tracking Processing Tool: An Exploratory Study.

Source check: Source checked. Not independent scientific confirmation.

Finding: A DeepLabCut/RGB-D pipeline generated 3D infant movement metrics in home recordings from eight infants; seven of eight showed patterns the authors judged consistent with expectations relative to two clinicians’ qualitative concern ratings. This is exploratory feasibility evidence, not validated screening or diagnosis. External replication and prospective utility were not established in the reviewed material.

Technology: Methods §§2.2–2.4: individualized DeepLabCut pose tracking, paired with depth to form 3D trajectories and movement metrics.

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
Methods §§2.2–2.4: individualized DeepLabCut pose tracking, paired with depth to form 3D trajectories and movement metrics.
Test and result
A DeepLabCut/RGB-D pipeline generated 3D infant movement metrics in home recordings from eight infants; seven of eight showed patterns the authors judged consistent with expectations relative to two clinicians’ qualitative concern ratings. This is exploratory feasibility evidence, not validated screening or diagnosis.
Limitation / still unknown
External replication, causal effects and prospective utility were not established in the reviewed material.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://iris.polito.it/retrieve/61f3ac46-06df-47b5-b82a-61b2bde437bd/Cereatti-Characterization.1.pdf

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Urine-output neural model shows retrospective discrimination of severe kidney-injury events

Source check: Source checked. Not independent scientific confirmation.

Finding: A convolutional neural network analyzed urine-output trends in 35,573 ICU patients to estimate later severe kidney-injury events under a restrictive urine-plus-creatinine definition. Patient-separated retrospective testing reported AUC 0.89, above logistic regression. Clinical prevention, calibrated alarm yield and prospective bedside benefit were not established.

Technology: Stacked and parallel convolutional neural networks with highway connections for urine-output time series

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
The CNN learned temporal urine-output features linked to later recorded severe-AKI labels and was compared with a logistic baseline. The admitted result is specific retrospective clinical discrimination, not recovery of a single urine-output cutoff or a validated intervention.
Test and result
The source pooled eICU and MIMIC-III records and separated patients across training, validation, calibration and test sets. The CNN used twelve-hour urine-output windows, with case series truncated six hours before the defined event. In the 7,080-patient test cohort, reported AUC was 0.89 versus 0.85 for logistic regression. The target required both creatinine increase and urine-output reduction; dialysis patients were excluded.
Limitation / still unknown
This was retrospective pooled-database evaluation, not a locked prospective clinical test. Urine output contributes to both model input and event definition; imputation and restrictive exclusions limit generalization to broader AKI. Reported likelihood ratios must not be read as positive predictive value or five true cases per six alarms. Individual-record timing, exact alarm yield and calibrated probabilities were not independently reconstructed. No reduced kidney injury, mortality or treatment benefit was demonstrated.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://link.springer.com/article/10.1007/s40620-021-01046-6

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 21 Nov 2026. Review scope: Read actual primary AKI target/eligibility, urine imputation and time truncation, patient-separated training/validation/calibration/test, CNN architecture and evaluation Methods plus clinical discrimination Results. Alarm likelihood-ratio interpretation checked and excluded from represented yield claims; individual-record timing and supplement not reconstructed. Exact original article front DOI/title and fresh PubMed primary relationship metadata checked..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open the interactive profile for the timeline, or browse all organizations and model families.