Truace
Subscribe
The Index recomputed live from the record

What the evidence says.What the public feels.

The record holds 932 sourced gains and 770 sourced problems, averaging 68 and 66 on the index score. Readers have logged 22 public signals on the Pulse, which is kept apart and never counted as evidence.

932 gains
770 problems
Every sourced claim in the Index, one square each, shaded by the strength of its evidence. High Moderate Emerging

Ranks distinct AI gain and problem claims from the published record. Scores reward impact, independent source strength, scale, confidence, and recency.

1,702 results
Show filters and sorting

Download every matching row, not just this page:Export CSVExport JSON

AI gains · 932

97
HealthStableModerate evidence · 1 source

Machine learning models trained on clinical, radiomics and dosiomics features predicted radionecrosis after stereotactic radiotherapy for brain metastases with ROC-AUC up to 80%, offering decision support that may improve patient-specific treatment and reduce radiotherapy-induced toxicity severity.

In a study published September 13, 2026, investigators applied eleven machine learning models to predict late radionecrosis after stereotactic radiotherapy for brain metastases. Using data from 37 patients with 113 lesions, where radionecrosis occurred in 18.6% of lesions, they tested four feature combinations of clinical, radiomics and dosiomics data after a 70:30 split and feature selection and balancing steps.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
95

Updated Sep 15, 2026 · TRV-2026-1098

74Index score
98
HealthStableModerate evidence · 1 source

A ResNet-18-based model trained on neonatal ICU radiographs detected pneumothorax on supine chest radiographs with AUC 0.975, 87.6% sensitivity and 95.3% specificity in the test set.

Researchers developed and evaluated a ResNet-18-based deep learning model to detect pneumothorax on supine neonatal chest radiographs using a retrospective dataset from a single NICU between 2011 and 2024. The model was pre-trained on normal adult radiographs and tested on held-out neonatal cases with patient-level splitting and a fixed threshold from cross-validation.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
95

Updated Sep 14, 2026 · TRV-2026-1085

74Index score
99
HealthStableModerate evidence · 1 source

A 1-D CNN with sharpness-aware minimization applied to Raman spectra distinguished unstimulated, activated, and exhausted T cells with >97% accuracy and quantified exhaustion percentage in heterogeneous populations with R2 = 1.

On September 12, 2026, a peer-reviewed study in ACS Sensors reported a label-free assay using Raman spectroscopy plus a 1-D convolutional neural network with sharpness aware minimization to distinguish T cell states directly from culture while preserving viability. The system discriminated unstimulated, activated, and exhausted T cells with >97% accuracy across three donors and hardware setups.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
95

Updated Sep 14, 2026 · TRV-2026-1084

74Index score
100
HealthStableModerate evidence · 1 source

Explainable ML models achieved near-perfect discrimination for breast cancer diagnosis on cytology data, with top models reaching 0.996 AUC and 98.25% accuracy, supporting use in resource-constrained diagnostic workflows.

On September 11, 2026, a peer-reviewed study in PLOS Digital Health reported an explainable AI framework for breast cancer diagnosis designed for underserved settings. Using 569 fine-needle aspirate specimens from the Wisconsin dataset, the authors benchmarked eight supervised classifiers with 10-fold cross-validation and a hold-out test set, then applied SHAP analysis to surface global and individual-level feature contributions.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
95

Updated Sep 13, 2026 · TRV-2026-1071

74Index score

AI problems · 770

97
HealthStableModerate evidence · 1 source

Chi-squared selection with a neural network achieved the highest mean accuracy (74.78%; AUC 0.77).

Nottingham histological grading is central to breast cancer prognosis and treatment planning, but conventional pathological assessment is labor-intensive and subject to inter-observer variability. Radiomics and machine learning may support noninvasive preoperative grade prediction.

Impact 30%
63
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
91

Updated Aug 23, 2026 · TRV-2026-0858

72Index score
98
HealthStableModerate evidence · 1 source

Use of generative AI in healthcare introduces challenges including lack of professional expertise in decision making, risk to patient data privacy, difficulty integrating with existing healthcare systems, and data bias.

This January 2024 IEEE Access review surveys generative AI in healthcare, describing models including ChatGPT, DALL-E, Bard, and seven healthcare-customized LLMs such as Med-PaLM, BioGPT, and DeepHealth. It catalogs applications from medical imaging and drug discovery to personalized treatment, simulation and training, clinical trial optimization, and medical chatbots, and details four real-world scenarios employing GAI: visual snow syndrome diagnosis, molecular drug optimization, medical education, and dentistry.

Impact 30%
49
Evidence 25%
95
Scale 20%
60
Confidence 15%
87
Recency 10%
89

Updated Aug 14, 2026 · TRV-2026-0760

72Index score
99
Media & ArtsStableModerate evidence · 1 source

Tacit rules are negatively associated with AI autonomy feasibility, constraining full replacement to 2.7% of tasks and leaving 11.0% AI immune in cultural and creative occupations.

Researchers analyzed 593 tasks across 126 occupations in the cultural and creative industries using GPT-4 generated synthetic annotations of Australian Skills Classification descriptions. They measured cognitive and behavioural rules and estimated AI autonomy feasibility and efficiency potential to map where human, AI, or hybrid carriers fit.

Impact 30%
63
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
88

Updated Aug 8, 2026 · TRV-2026-0693

72Index score
100
HealthStableModerate evidence · 1 source

Patients classified as Severe profile with early onset, 73.3% co-occurring psychiatric conditions and intense craving had poorer 3-month outcomes with only 36.7% abstinence and steep return to use.

In a retrospective study of 102 patients at a tertiary care center in India, researchers used k-means clustering on eight biopsychosocial baseline variables to derive three AUD profiles. By the August 2026 publication date, they reported Late-Onset, High-Functioning, and Severe groups with differing 3-month abstinence rates corroborated by GGT levels and bootstrap-assessed cluster stability.

Impact 30%
63
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
88

Updated Aug 8, 2026 · TRV-2026-0687

72Index score

Recomputed live from the record · Oct 11, 2026, 10:57 AM