Does your dermatology classifier know what it doesn’t know? Detecting the long-tail of unseen conditions

Roy, Abhijit Guha; Ren, Jie; Azizi, Shekoofeh; Loh, Aaron; Natarajan, Vivek; Mustafa, Basil; Pawlowski, Nick; Freyberg, Jan; Liu, Yuan; Beaver, Zach; Vo, Nam; Bui, Peggy; Winter, Samantha; MacWilliams, Patricia; Corrado, Greg S.; Telang, Umesh; Liu, Yun; Cemgil, Taylan; Karthikesalingam, Alan; Lakshminarayanan, Balaji; Winkens, Jim

doi:10.1016/j.media.2021.102274

Cited by 36 publications

(21 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Approaches may build on previous work that combines language models and knowledge graphs 25,26 to reason step-by-step about surgical tasks. Additionally, GMAI deployed in surgical settings will probably face unusual clinical phenomena that cannot be included during model development, owing to their rarity, a challenge known as the long tail of unseen conditions 27 . Medical reasoning abilities will be crucial for both detecting previously unseen outliers and explaining them, as exemplified in Fig.…”

Section: Grounded Radiology Reports Gmai Enables a New Generation Of ...mentioning

confidence: 99%

Foundation models for generalist medical artificial intelligence

Moor

Banerjee

Abad

et al. 2023

Nature

389

125

View full text Add to dashboard Cite

The exceptionally rapid development of highly flexible, reusable artificial intelligence (AI) models is likely to usher in newfound capabilities in medicine. We propose a new paradigm for medical AI, which we refer to as generalist medical AI (GMAI). GMAI models will be capable of carrying out a diverse set of tasks using very little or no task-specific labelled data. Built through self-supervision on large, diverse datasets, GMAI will flexibly interpret different combinations of medical modalities, including data from imaging, electronic health records, laboratory results, genomics, graphs or medical text. Models will in turn produce expressive outputs such as free-text explanations, spoken recommendations or image annotations that demonstrate advanced medical reasoning abilities. Here we identify a set of high-impact potential applications for GMAI and lay out specific technical capabilities and training datasets necessary to enable them. We expect that GMAI-enabled applications will challenge current strategies for regulating and validating AI devices for medicine and will shift practices associated with the collection of large medical datasets.

show abstract

Section: Grounded Radiology Reports Gmai Enables a New Generation Of ...mentioning

confidence: 99%

Foundation models for generalist medical artificial intelligence

Moor

Banerjee

Abad

et al. 2023

Nature

389

125

View full text Add to dashboard Cite

show abstract

“…Promisingly, ISIC provides a template of the large, open‐access datasets that have driven improvement of ML in other fields, 22,49–51 and it continues to increase in size. Technological advances may also help the robustness of ML model performance, such as abstaining from predictions in rare or unknown disease classes 42 …”

Section: Discussionmentioning

confidence: 99%

“…Thirdly, although we used a low threshold for considering a journal ‘medical’, some studies could be missed from non‐medical journal sources. However, in studies identified in non‐medical journals that considered rare disease classes, the focus was on methodological modifications rather than clinical performance 42,53,54 . Finally, some studies may have reported the number of training images used after performing data augmentation, and thus, image counts may be inaccurate.…”

Section: Discussionmentioning

confidence: 99%

Determining the clinical applicability of machine learning models through assessment of reporting across skin phototypes and rarer skin cancer types: A systematic review

Steele

Tan

Olabi

et al. 2023

Acad Dermatol Venereol

View full text Add to dashboard Cite

Machine learning (ML) models for skin cancer recognition may have variable performance across different skin phototypes and skin cancer types. Overall performance metrics alone are insufficient to detect poor subgroup performance. We aimed (1) to assess whether studies of ML models reported results separately for different skin phototypes and rarer skin cancers, and (2) to graphically represent the skin cancer training datasets used by current ML models. In this systematic review, we searched PubMed, Embase and CENTRAL. We included all studies in medical journals assessing an ML technique for skin cancer diagnosis that used clinical or dermoscopic images from 1 January 2012 to 22 September 2021. No language restrictions were applied. We considered rarer skin cancers to be skin cancers other than pigmented melanoma, basal cell carcinoma and squamous cell carcinoma. We identified 114 studies for inclusion. Rarer skin cancers were included by 8/114 studies (7.0%), and results for a rarer skin cancer were reported separately in 1/114 studies (0.9%). Performance was reported across all skin phototypes in 1/114 studies (0.9%), but performance was uncertain in skin phototypes I and VI from minimal representation of the skin phototypes in the test dataset (9/3756 and 1/3756, respectively). For training datasets, although public datasets were most frequently used, with the most widely used being the International Skin Imaging Collaboration (ISIC) archive (65/114 studies, 57.0%), the largest datasets were private. Our review identified that most ML models did not report performance separately for rarer skin cancers and different skin phototypes. A degree of variability in ML model performance across subgroups is expected, but the current lack of transparency is not justifiable and risks models being used inappropriately in populations in whom accuracy is low.

show abstract

“…Describe if images with classes that are OOD were included in the study test set, and report findings. 45 If images with OOD classes were not assessed, explain the drawbacks to clinical application (ie, undefined behavior when presented with classes outside of those studied). In some cases, OOD data may be subtle-for example, beyond classes not represented in training data, OOD may include unique combinations of other characteristics, such as clinical site, camera used, lighting, and patient demographics, of which some combinations may be underrepresented in algorithm training data.…”

Section: Meaningmentioning

confidence: 99%

“…For example, if an algorithm is trained to differentiate nevi vs melanomas, any image showing a diagnosis outside of nevi and melanomas would be OOD. Describe if images with classes that are OOD were included in the study test set, and report findings . If images with OOD classes were not assessed, explain the drawbacks to clinical application (ie, undefined behavior when presented with classes outside of those studied).…”

Section: Recommendationsmentioning

confidence: 99%