Bibliographic record
Abstract
Background Recently, deep convolutional neural networks (DCNNs) became of interest as decision support systems for dermoscopic and clinical analysis of skin diseases. Application of artificial intelligence in teledermatology (TD) has been recently reported in several studies as a tool for augmented intelligence. Objective In this session, a critical discussion of the opportunities, limitations, and risks of AI in TD will be presented with special attention to recent published studies. Methods We reviewed the literature in PubMed and EMBASE databases in the period of January 2018 to November 2021 with the search terms of dermatology, skin cancer, deep learning, and AI (review of the Regulation of Medical Devices, EU 2017/745). Results A clear definition of the clinical use of AI in TD has to be considered: primary TD from patients to nurses, primary care physicians or general dermatologists; secondary TD from primary care physicians or nurses to dermatologists; or tertiary TD from dermatologists to hospital dermatologists. In some health models of TD for nurses or primary care physicians, AI assistance can lower the rates of recommending a biopsy or specialist referral, increase self-reported diagnostic confidence, and help to achieve higher diagnostic agreement rates (with dermatologists) in nonreferred cases. The main limitations of the use of AI in TD are the lack of large longitudinal studies, the lack of interpretability of the CNN, biases in the databases and unrepresented dermatological conditions for training, limited representation of different ethnicities, standardization of clinical information and of the images, liability, and privacy issues. How to implement the concept of augmented intelligence in clinical practice with referral TD consultations including structured clinical information and good-quality images will need further research and education among end users. Even if the interface is used for either store-and-forward or live TD, interactive TD is, in principle, straightforward for AI systems, and different TD modalities have particular technological requirements that can reduce their efficacy. Finally, AI systems in TD are under the umbrella of medical device regulatory frames, and specific certification is compulsory. This regulation has the benefit of assuring the quality of the new AI systems and diminishing their risks, but it can simultaneously delay the incorporation of AI tools in clinical practice. Conclusions AI has the potential to improve the results of the technology in different aspects in multiple modalities of TD. However, the evidence is weak, and several barriers and limitations have to be resolved for their integration in clinical practice. Conflicts of Interest None declared.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.031 | 0.094 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.001 | 0.006 |
| Scholarly communication | 0.005 | 0.006 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.004 | 0.004 |
| Insufficient payload (model declined to judge) | 0.006 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".