Assessment of Deep Learning Using Nonimaging Information and Sequential Medical Records to Develop a Prediction Model for Nonmelanoma Skin Cancer

Artificial intelligence and skin cancer

Artificial intelligence is poised to rapidly reshape many fields, including that of skin cancer screening and diagnosis, both as a disruptive and assistive technology. Together with the collection and availability of large medical data sets, artificial intelligence will become a powerful tool that can be leveraged by physicians in their diagnoses and treatment plans for patients. This comprehensive review focuses on current progress toward AI applications for patients, primary care providers, dermatologists, and dermatopathologists, explores the diverse applications of image and molecular processing for skin cancer, and highlights AI's potential for patient self-screening and improving diagnostic accuracy for non-dermatologists. We additionally delve into the challenges and barriers to clinical implementation, paths forward for implementation and areas of active research.

Applications of Machine Learning to Diagnosis of Parkinson's Disease

BACKGROUND

Accurate diagnosis of Parkinson's disease (PD) is challenging due to its diverse manifestations. Machine learning (ML) algorithms can improve diagnostic precision, but their generalizability across medical centers in China is underexplored.

OBJECTIVE

To assess the accuracy of an ML algorithm for PD diagnosis, trained and tested on data from different medical centers in China.

METHODS

A total of 1656 participants were included, with 1028 from Beijing (training set) and 628 from Fuzhou (external validation set). Models were trained using the least absolute shrinkage and selection operator-logistic regression (LASSO-LR), decision tree (DT), random forest (RF), eXtreme gradient boosting (XGboost), support vector machine (SVM), and k-nearest neighbor (KNN) techniques. Hyperparameters were optimized using five-fold cross-validation and grid search techniques. Model performance was evaluated using the area under the curve (AUC) of the receiver operating characteristic (ROC) curve, accuracy, sensitivity (recall), specificity, precision, and F1 score. Variable importance was assessed for all models.

RESULTS

SVM demonstrated the best differentiation between healthy controls (HCs) and PD patients (AUC: 0.928, 95% CI: 0.908-0.947; accuracy: 0.844, 95% CI: 0.814-0.871; sensitivity: 0.826, 95% CI: 0.786-0.866; specificity: 0.861, 95% CI: 0.820-0.898; precision: 0.849, 95% CI: 0.807-0.891; F1 score: 0.837, 95% CI: 0.803-0.868) in the validation set. Constipation, olfactory decline, and daytime somnolence significantly influenced predictability.

CONCLUSION

We identified multiple pivotal variables and SVM as a precise and clinician-friendly ML algorithm for prediction of PD in Chinese patients.

Machine Learning-Based Prediction of Atrial Fibrillation Risk Using Electronic Medical Records in Older Aged Patients

Atrial fibrillation (AF) is an independent risk factor that increases the risk of stroke 5-fold. The purpose of our study was to develop a 1-year new-onset AF predictive model by machine learning based on 3-year medical information without electrocardiograms in our database to identify AF risk in older aged patients. We developed the predictive model according to the Taipei Medical University clinical research database electronic medical records, including diagnostic codes, medications, and laboratory data. Decision tree, support vector machine, logistic regression, and random forest algorithms were chosen for the analysis. A total of 2,138 participants (1,028 women [48.1%]; mean [standard deviation] age 78.8 [6.8] years) with AF and 8,552 random controls (after the matching process) without AF (4,112 women [48.1%]; mean [standard deviation] age 78.8 [6.8] years) were included in the model. The 1-year new-onset AF risk prediction model based on the random forest algorithm using medication and diagnostic information, along with specific laboratory data, attained an area under the receiver operating characteristic of 0.74, whereas the specificity was 98.7%. Machine learning-based model focusing on the older aged patients could offer acceptable discrimination in differentiating the risk of incident AF in the next year. In conclusion, a targeted screening approach using multidimensional informatics in the electronic medical records could result in a clinical choice with efficacy for prediction of the incident AF risk in older aged patients.

Overinterpretation of findings in machine learning prediction model studies in oncology: a systematic review

OBJECTIVES

In biomedical research, spin is the overinterpretation of findings, and it is a growing concern. To date, the presence of spin has not been evaluated in prognostic model research in oncology, including studies developing and validating models for individualized risk prediction.

STUDY DESIGN AND SETTING

We conducted a systematic review, searching MEDLINE and EMBASE for oncology-related studies that developed and validated a prognostic model using machine learning published between 1st January, 2019, and 5th September, 2019. We used existing spin frameworks and described areas of highly suggestive spin practices.

RESULTS

We included 62 publications (including 152 developed models; 37 validated models). Reporting was inconsistent between methods and the results in 27% of studies due to additional analysis and selective reporting. Thirty-two studies (out of 36 applicable studies) reported comparisons between developed models in their discussion and predominantly used discrimination measures to support their claims (78%). Thirty-five studies (56%) used an overly strong or leading word in their title, abstract, results, discussion, or conclusion.

CONCLUSION

The potential for spin needs to be considered when reading, interpreting, and using studies that developed and validated prognostic models in oncology. Researchers should carefully report their prognostic model research using words that reflect their actual results and strength of evidence.

Multi-site cross-organ calibrated deep learning (MuSClD): Automated diagnosis of non-melanoma skin cancer

Although deep learning (DL) has demonstrated impressive diagnostic performance for a variety of computational pathology tasks, this performance often markedly deteriorates on whole slide images (WSI) generated at external test sites. This phenomenon is due in part to domain shift, wherein differences in test-site pre-analytical variables (e.g., slide scanner, staining procedure) result in WSI with notably different visual presentations compared to training data. To ameliorate pre-analytic variances, approaches such as CycleGAN can be used to calibrate visual properties of images between sites, with the intent of improving DL classifier generalizability. In this work, we present a new approach termed Multi-Site Cross-Organ Calibration based Deep Learning (MuSClD) that employs WSIs of an off-target organ for calibration created at the same site as the on-target organ, based off the assumption that cross-organ slides are subjected to a common set of pre-analytical sources of variance. We demonstrate that by using an off-target organ from the test site to calibrate training data, the domain shift between training and testing data can be mitigated. Importantly, this strategy uniquely guards against potential data leakage introduced during calibration, wherein information only available in the testing data is imparted on the training data. We evaluate MuSClD in the context of the automated diagnosis of non-melanoma skin cancer (NMSC). Specifically, we evaluated MuSClD for identifying and distinguishing (a) basal cell carcinoma (BCC), (b) in-situ squamous cell carcinomas (SCC-In Situ), and (c) invasive squamous cell carcinomas (SCC-Invasive), using an Australian (training, n = 85) and a Swiss (held-out testing, n = 352) cohort. Our experiments reveal that MuSCID reduces the Wasserstein distances between sites in terms of color, contrast, and brightness metrics, without imparting noticeable artifacts to training data. The NMSC-subtyping performance is statistically improved as a result of MuSCID in terms of one-vs. rest AUC: BCC (0.92 vs 0.87, p = 0.01), SCC-In Situ (0.87 vs 0.73, p = 0.15) and SCC-Invasive (0.92 vs 0.82, p = 1e-5). Compared to baseline NMSC-subtyping with no calibration, the internal validation results of MuSClD (BCC (0.98), SCC-In Situ (0.92), and SCC-Invasive (0.97)) suggest that while domain shift indeed degrades classification performance, our on-target calibration using off-target tissue can safely compensate for pre-analytical variabilities, while improving the robustness of the model.

Use of convolutional neural networks in skin lesion analysis using real world image and non-image data

Background

Understanding performance of convolutional neural networks (CNNs) for binary (benign vs. malignant) lesion classification based on real world images is important for developing a meaningful clinical decision support (CDS) tool.

Methods

We developed a CNN based on real world smartphone images with histopathological ground truth and tested the utility of structured electronic health record (EHR) data on model performance. Model accuracy was compared against three board-certified dermatologists for clinical validity.

Results

At a classification threshold of 0.5, the sensitivity was 79 vs. 77 vs. 72%, and specificity was 64 vs. 65 vs. 57% for image-alone vs. combined image and clinical data vs. clinical data-alone models, respectively. The PPV was 68 vs. 69 vs. 62%, AUC was 0.79 vs. 0.79 vs. 0.69, and AP was 0.78 vs. 0.79 vs. 0.64 for image-alone vs. combined data vs. clinical data-alone models. Older age, male sex, and number of prior dermatology visits were important positive predictors for malignancy in the clinical data-alone model.

Conclusion

Additional clinical data did not significantly improve CNN image model performance. Model accuracy for predicting malignant lesions was comparable to dermatologists (model: 71.31% vs. 3 dermatologists: 77.87, 69.88, and 71.93%), validating clinical utility. Prospective validation of the model in primary care setting will enhance understanding of the model’s clinical utility.

Major areas of interest of artificial intelligence research applied to health care administrative data: a scoping review

Introduction: The ongoing collection of large medical data has created conditions for application of artificial intelligence (AI) in research. This scoping review aimed to identify major areas of interest of AI applied to health care administrative data.

Methods: The search was performed in seven databases: Medline, Embase, CINAHL, Web of science, IEEE, ICM digital library, and Compendex. We included articles published between January 2001 and March 2021, that described research with AI applied to medical diagnostics, pharmacotherapy, and health outcomes data. We screened the full text content and used natural language processing to automatically extract health areas of interest, principal AI methods, and names of medications.

Results: Out of 14,864 articles, 343 were included. We determined ten areas of interest, the most common being health diagnostic or treatment outcome prediction (32%); representation of medical data, clinical pathways, and data temporality (i.e., transformation of raw medical data into compact and analysis-friendly format) (22%); and adverse drug effects, drug-drug interactions, and medication cascades (15%). Less attention has been devoted to areas such as health effects of polypharmacy (1%); and reinforcement learning (1%). The most common AI methods were decision trees, cluster analysis, random forests, and support vector machines. Most frequently mentioned medications included insulin, metformin, vitamins, acetaminophen, and heparin.

Conclusions: The scoping review revealed the potential of AI application to health-related studies. However, several areas of interest in pharmacoepidemiology are sparsely reported, and the lack of details in studies related to pharmacotherapy suggests that AI could be used more optimally in pharmacoepidemiologic research.

Ability to Predict Melanoma Within 5 Years Using Registry Data and a Convolutional Neural Network: A Proof of Concept Study

Research relating to machine learning algorithms, including convolutional neural networks, has increased during the past 5 years. The aim of this pilot study was to investigate how accurately a convolutional neural network, trained on Swedish registry data, could perform in predicting cutaneous invasive and in situ melanoma (CMM) within 5 years. A cohort of 1,208,393 individuals was used. Registry data ranged from 4 July 2005 to 31 December 2011, predicting CMM between 1 January 2012 and 31 December 2016. A convolutional neural network with one-dimensional convolutions with respect to time was trained using healthcare databases and registers. The algorithm was trained on 23,886 individuals. Validation was performed on a holdout validation set including 6,000 individuals. After training and validation, the convolutional neural network was evaluated on a test set (1,000 individuals with an CMM occurring within 5 years and 5,000 without). The area under the receiver-operating characteristic curve was 0.59 (95% confidence interval (95% CI) 0.57–0.61). The point on the receiver-operating characteristic curve where sensitivity equalled specificity had a value of 56% (sensitivity 95% CI 53–60% and specificity 95% CI 55–58%). Albeit at an early stage, this pilot investigation demonstrates potential usefulness for machine learning algorithms in predicting melanoma risk.

Causal Association between Chronic Kidney Disease and Risk of 19 Site-Specific Cancers: A Mendelian Randomization Study

BACKGROUND

Results of previous observational studies examining the risk of cancer among patients with chronic kidney disease (CKD) are conflicting. We here explore the causal relationship between estimated glomerular filtration rate (eGFR) and albuminuria, two principal measurements of CKD, and 19 site-specific cancers using Mendelian randomization (MR) analysis.

METHODS

Single-nucleotide polymorphisms reported to be strongly correlated with eGFR and albuminuria in recent large genome-wide association studies were used as instrumental variables to investigate the causal relationship with cancer using summary-level statistics from several cancer-specific consortia, as well as data of 347,408 participants in the UK Biobank and 260,405 participants in the FinnGen.

RESULTS

Our data showed that impaired kidney function was associated with higher odds of leukemia [OR = 1.23; 95% confidence interval (CI), 1.06-1.43; P = 0.007], cervical cancer (OR = 1.22; 95% CI, 1.04-1.43; P = 0.017), and female renal cell carcinoma (OR = 1.4; 95% CI, 1.12-1.77; P = 0.004), per 10% decrease in eGFR. The ORs were 1.21 (95% CI, 1.07-1.36; P = 0.002) for colorectal cancer and 0.76 (95% CI, 0.62-0.92; P = 0.006) for non-Hodgkin lymphoma, per doubling odds of albuminuria. In multivariable MR, effect sizes of eGFR-cervical cancer remained strong after adjusting for confounders.

CONCLUSIONS

The current study indicates that progression of CKD contributes to carcinogenesis of renal cell carcinoma, leukemia, cervical, and colorectal cancer.

IMPACT

The potential association of kidney function and albuminuria with certain cancers warrants further investigation in order to provide appropriate recommendations regarding cancer screening among patients with CKD.

Assessment of the Predictive Value of Spectrophotometric Skin Color Parameters and Environmental and Behavioral Factors in Estimating the Risk of Skin Cancer: A Case-Control Study

Keratinocyte carcinomas are the most common cancers with different etiological risk factors. The aim of this study was to assess the predictive value of spectrophotometric parameters of skin color in correlation with environmental/behavioral factors to estimate the risk of skin cancer. The case-control study involved 389 patients. The analysis was performed on the training group to build a predictive model and on the testing group to check the quality of the designed model. Area under the curve based on the spectrophotometric skin parameters varied from 0.536 to 0.674. A statistically significant improvement of the area under curve was achieved by adding the number of sunburns for some models. The best single spectrophotometric measurement for estimating skin cancer is the skin melanin index measured on the arm or buttock. Spectrophotometric skin parameters are not very strong but are essential elements of models for estimating the risk of skin cancer. The most important environmental/behavioral factor seems to be the number of sunburns, but not the total exposure to ultraviolet radiation or usage of photoprotectors. Some other pigmentation predictors should be taken into account when creating new models, especially those that can be easily measured in objective and repeatable way. Spectrophotometric measurements can be employed as quick screening skin examination method.

Methodological conduct of prognostic prediction models developed using machine learning in oncology: a systematic review

BACKGROUND

Describe and evaluate the methodological conduct of prognostic prediction models developed using machine learning methods in oncology.

METHODS

We conducted a systematic review in MEDLINE and Embase between 01/01/2019 and 05/09/2019, for studies developing a prognostic prediction model using machine learning methods in oncology. We used the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) statement, Prediction model Risk Of Bias ASsessment Tool (PROBAST) and CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS) to assess the methodological conduct of included publications. Results were summarised by modelling type: regression-, non-regression-based and ensemble machine learning models.

RESULTS

Sixty-two publications met inclusion criteria developing 152 models across all publications. Forty-two models were regression-based, 71 were non-regression-based and 39 were ensemble models. A median of 647 individuals (IQR: 203 to 4059) and 195 events (IQR: 38 to 1269) were used for model development, and 553 individuals (IQR: 69 to 3069) and 50 events (IQR: 17.5 to 326.5) for model validation. A higher number of events per predictor was used for developing regression-based models (median: 8, IQR: 7.1 to 23.5), compared to alternative machine learning (median: 3.4, IQR: 1.1 to 19.1) and ensemble models (median: 1.7, IQR: 1.1 to 6). Sample size was rarely justified (n = 5/62; 8%). Some or all continuous predictors were categorised before modelling in 24 studies (39%). 46% (n = 24/62) of models reporting predictor selection before modelling used univariable analyses, and common method across all modelling types. Ten out of 24 models for time-to-event outcomes accounted for censoring (42%). A split sample approach was the most popular method for internal validation (n = 25/62, 40%). Calibration was reported in 11 studies. Less than half of models were reported or made available.

CONCLUSIONS

The methodological conduct of machine learning based clinical prediction models is poor. Guidance is urgently needed, with increased awareness and education of minimum prediction modelling standards. Particular focus is needed on sample size estimation, development and validation analysis methods, and ensuring the model is available for independent validation, to improve quality of machine learning based clinical prediction models.

Synthesizing time-series wound prognosis factors from electronic medical records using generative adversarial networks

Wound prognostic models not only provide an estimate of wound healing time to motivate patients to follow up their treatments but also can help clinicians to decide whether to use a standard care or adjuvant therapies and to assist them with designing clinical trials. However, collecting prognosis factors from Electronic Medical Records (EMR) of patients is challenging due to privacy, sensitivity, and confidentiality. In this study, we developed time series medical generative adversarial networks (GANs) to generate synthetic wound prognosis factors using very limited information collected during routine care in a specialized wound care facility. The generated prognosis variables are used in developing a predictive model for chronic wound healing trajectory. Our novel medical GAN can produce both continuous and categorical features from EMR. Moreover, we applied temporal information to our model by considering data collected from the weekly follow-ups of patients. Conditional training strategies were utilized to enhance training and generate classified data in terms of healing or non-healing. The ability of the proposed model to generate realistic EMR data was evaluated by TSTR (test on the synthetic, train on the real), discriminative accuracy, and visualization. We utilized samples generated by our proposed GAN in training a prognosis model to demonstrate its real-life application. Using the generated samples in training predictive models improved the classification accuracy by 6.66-10.01% compared to the previous EMR-GAN. Additionally, the suggested prognosis classifier has achieved the area under the curve (AUC) of 0.875, 0.810, and 0.647 when training the network using data from the first three visits, first two visits, and first visit, respectively. These results indicate a significant improvement in wound healing prediction compared to the previous prognosis models.

Predicting the risk of cancer in adults using supervised machine learning: a scoping review

OBJECTIVES

The purpose of this scoping review is to: (1) identify existing supervised machine learning (ML) approaches on the prediction of cancer in asymptomatic adults; (2) to compare the performance of ML models with each other and (3) to identify potential gaps in research.

DESIGN

Scoping review using the population, concept and context approach.

SEARCH STRATEGY

PubMed search engine was used from inception to 10 November 2020 to identify literature meeting following inclusion criteria: (1) a general adult (≥18 years) population, either sex, asymptomatic (population); (2) any study using ML techniques to derive predictive models for future cancer risk using clinical and/or demographic and/or basic laboratory data (concept) and (3) original research articles conducted in all settings in any region of the world (context).

RESULTS

The search returned 627 unique articles, of which 580 articles were excluded because they did not meet the inclusion criteria, were duplicates or were related to benign neoplasm. Full-text reviews were conducted for 47 articles and a final set of 10 articles were included in this scoping review. These 10 very heterogeneous studies used ML to predict future cancer risk in asymptomatic individuals. All studies reported area under the receiver operating characteristics curve (AUC) values as metrics of model performance, but no study reported measures of model calibration.

CONCLUSIONS

Research gaps that must be addressed in order to deliver validated ML-based models to assist clinical decision-making include: (1) establishing model generalisability through validation in independent cohorts, including those from low-income and middle-income countries; (2) establishing models for all cancer types; (3) thorough comparisons of ML models with best available clinical tools to ensure transparency of their potential clinical utility; (4) reporting of model calibration performance and (5) comparisons of different methods on the same cohort to reveal important information about model generalisability and performance.

A Deep Learning Model to Predict Knee Osteoarthritis Based on Nonimage Longitudinal Medical Record

PURPOSE

To develop deep learning model (Deep-KOA) that can predict the risk of knee osteoarthritis (KOA) within the next year by using the previous three years nonimage-based electronic medical record (EMR) data.

PATIENTS AND METHODS

We randomly selected information of two million patients from the Taiwan National Health Insurance Research Database (NHIRD) from January 1, 1999 to December 31, 2013. During the study period, 132,594 patients were diagnosed with KOA, while 1,068,464 patients without KOA were chosen randomly as control. We constructed a feature matrix by using the three-year history of sequential diagnoses, drug prescriptions, age, and sex. Deep learning methods of convolutional neural network (CNN) and artificial neural network (ANN) were used together to develop a risk prediction model. We used the area under the receiver operating characteristic (AUROC), sensitivity, specificity, and precision to evaluate the performance of Deep-KOA. Then, we explored the important features using stepwise feature selection.

RESULTS

This study included 132,594 KOA patients, 83,111 females (62.68%), 49,483 males (37.32%), mean age 64.2 years, and 1,068,464 non-KOA patients, 545,902 females (51.09%), 522,562 males (48.91%), mean age 51.00 years. The Deep-KOA achieved an overall AUROC, sensitivity, specificity, and precision of 0.97, 0.89, 0.93, and 0.80 respectively. The discriminative analysis of Deep-KOA showed important features from several diseases such as disorders of the eye and adnexa, acute respiratory infection, other metabolic and immunity disorders, and diseases of the musculoskeletal and connective tissue. Age and sex were not found as the most discriminative features, with AUROC of 0.9593 (-0.76% loss) and 0.9644 (-0.25% loss) respectively. Whereas medications including antacid, cough suppressant, and expectorants were identified as discriminative features.

CONCLUSION

Deep-KOA was developed to predict the risk of KOA within one year earlier, which may provide clues for clinical decision support systems to target patients with high risk of KOA to get precision prevention program.

Cutaneous Human Papillomaviruses and the Risk of Keratinocyte Carcinomas

Cutaneous human papillomavirus (cuHPV) infections may be novel targets for skin cancer prevention and treatment, but critical information regarding the development of virus-positive skin cancers following cuHPV infection has been lacking. In this study, baseline cuHPV infection was measured by serology and viral DNA detection in eyebrow hairs (EBH) and forearm skin swabs (SSW) among 1,008 individuals undergoing routine skin cancer screening exams and followed for incidence of basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cuSCC). Baseline β-HPV detection, particularly in SSW, significantly predicted cuSCC (HR = 4.32; 95% confidence interval, 1.00-18.66), whereas serologic evidence of past β-HPV infection was not associated with cuSCC. Less than 5% of baseline β-HPV types detected in SSW were present in subsequent cuSCC tumors, and cuHPV detected in SSW with higher mean fluorescence intensity values were more likely to be present in cuSCC compared with those with lower levels (P < 0.001). β-HPV-positive cuSCC occurred more often in areas of highly sun-damaged skin than did β-HPV-negative cuSCC. Overall, no clear patterns were observed between baseline β-HPV detection and subsequent development of BCC, or between baseline γ-HPV detection and either cuSCC or BCC. Collectively, these results demonstrate that β-HPV detection in SSW is a significant predictor of cuSCC risk, although evidence suggests only a small subset of cuSCC is etiologically linked to β-HPV infection. SIGNIFICANCE: β-HPV positivity may be a useful biomarker for identifying individuals who could benefit from increased screening or novel cutaneous squamous cell carcinoma prevention strategies.

Artificial Intelligence-Based Prediction of Lung Cancer Risk Using Nonimaging Electronic Medical Records: Deep Learning Approach

Background

Artificial intelligence approaches can integrate complex features and can be used to predict a patient’s risk of developing lung cancer, thereby decreasing the need for unnecessary and expensive diagnostic interventions.

Objective

The aim of this study was to use electronic medical records to prescreen patients who are at risk of developing lung cancer.

Methods

We randomly selected 2 million participants from the Taiwan National Health Insurance Research Database who received care between 1999 and 2013. We built a predictive lung cancer screening model with neural networks that were trained and validated using pre-2012 data, and we tested the model prospectively on post-2012 data. An age- and gender-matched subgroup that was 10 times larger than the original lung cancer group was used to assess the predictive power of the electronic medical record. Discrimination (area under the receiver operating characteristic curve [AUC]) and calibration analyses were performed.

Results

The analysis included 11,617 patients with lung cancer and 1,423,154 control patients. The model achieved AUCs of 0.90 for the overall population and 0.87 in patients ≥55 years of age. The AUC in the matched subgroup was 0.82. The positive predictive value was highest (14.3%) among people aged ≥55 years with a pre-existing history of lung disease.

Conclusions

Our model achieved excellent performance in predicting lung cancer within 1 year and has potential to be deployed for digital patient screening. Convolution neural networks facilitate the effective use of EMRs to identify individuals at high risk for developing lung cancer.

Integrating Patient Data Into Skin Cancer Classification Using Convolutional Neural Networks: Systematic Review

BACKGROUND

Recent years have been witnessing a substantial improvement in the accuracy of skin cancer classification using convolutional neural networks (CNNs). CNNs perform on par with or better than dermatologists with respect to the classification tasks of single images. However, in clinical practice, dermatologists also use other patient data beyond the visual aspects present in a digitized image, further increasing their diagnostic accuracy. Several pilot studies have recently investigated the effects of integrating different subtypes of patient data into CNN-based skin cancer classifiers.

OBJECTIVE

This systematic review focuses on the current research investigating the impact of merging information from image features and patient data on the performance of CNN-based skin cancer image classification. This study aims to explore the potential in this field of research by evaluating the types of patient data used, the ways in which the nonimage data are encoded and merged with the image features, and the impact of the integration on the classifier performance.

METHODS

Google Scholar, PubMed, MEDLINE, and ScienceDirect were screened for peer-reviewed studies published in English that dealt with the integration of patient data within a CNN-based skin cancer classification. The search terms skin cancer classification, convolutional neural network(s), deep learning, lesions, melanoma, metadata, clinical information, and patient data were combined.

RESULTS

A total of 11 publications fulfilled the inclusion criteria. All of them reported an overall improvement in different skin lesion classification tasks with patient data integration. The most commonly used patient data were age, sex, and lesion location. The patient data were mostly one-hot encoded. There were differences in the complexity that the encoded patient data were processed with regarding deep learning methods before and after fusing them with the image features for a combined classifier.

CONCLUSIONS

This study indicates the potential benefits of integrating patient data into CNN-based diagnostic algorithms. However, how exactly the individual patient data enhance classification performance, especially in the case of multiclass classification problems, is still unclear. Moreover, a substantial fraction of patient data used by dermatologists remains to be analyzed in the context of CNN-based skin cancer classification. Further exploratory analyses in this promising field may optimize patient data integration into CNN-based skin cancer diagnostics for patients' benefits.

An Application of Machine Learning to Etiological Diagnosis of Secondary Hypertension: Retrospective Study Using Electronic Medical Records

Background

Secondary hypertension is a kind of hypertension with a definite etiology and may be cured. Patients with suspected secondary hypertension can benefit from timely detection and treatment and, conversely, will have a higher risk of morbidity and mortality than those with primary hypertension.

Objective

The aim of this study was to develop and validate machine learning (ML) prediction models of common etiologies in patients with suspected secondary hypertension.

Methods

The analyzed data set was retrospectively extracted from electronic medical records of patients discharged from Fuwai Hospital between January 1, 2016, and June 30, 2019. A total of 7532 unique patients were included and divided into 2 data sets by time: 6302 patients in 2016-2018 as the training data set for model building and 1230 patients in 2019 as the validation data set for further evaluation. Extreme Gradient Boosting (XGBoost) was adopted to develop 5 models to predict 4 etiologies of secondary hypertension and occurrence of any of them (named as composite outcome), including renovascular hypertension (RVH), primary aldosteronism (PA), thyroid dysfunction, and aortic stenosis. Both univariate logistic analysis and Gini Impurity were used for feature selection. Grid search and 10-fold cross-validation were used to select the optimal hyperparameters for each model.

Results

Validation of the composite outcome prediction model showed good performance with an area under the receiver-operating characteristic curve (AUC) of 0.924 in the validation data set, while the 4 prediction models of RVH, PA, thyroid dysfunction, and aortic stenosis achieved AUC of 0.938, 0.965, 0.959, and 0.946, respectively, in the validation data set. A total of 79 clinical indicators were identified in all and finally used in our prediction models. The result of subgroup analysis on the composite outcome prediction model demonstrated high discrimination with AUCs all higher than 0.890 among all age groups of adults.

Conclusions

The ML prediction models in this study showed good performance in detecting 4 etiologies of patients with suspected secondary hypertension; thus, they may potentially facilitate clinical diagnosis decision making of secondary hypertension in an intelligent way.

Man against machine reloaded: performance of a market-approved convolutional neural network in classifying a broad spectrum of skin lesions in comparison with 96 dermatologists working under less artificial conditions

BACKGROUND

Convolutional neural networks (CNNs) efficiently differentiate skin lesions by image analysis. Studies comparing a market-approved CNN in a broad range of diagnoses to dermatologists working under less artificial conditions are lacking.

MATERIALS AND METHODS

One hundred cases of pigmented/non-pigmented skin cancers and benign lesions were used for a two-level reader study in 96 dermatologists (level I: dermoscopy only; level II: clinical close-up images, dermoscopy, and textual information). Additionally, dermoscopic images were classified by a CNN approved for the European market as a medical device (Moleanalyzer Pro, FotoFinder Systems, Bad Birnbach, Germany). Primary endpoints were the sensitivity and specificity of the CNN's dichotomous classification in comparison with the dermatologists' management decisions. Secondary endpoints included the dermatologists' diagnostic decisions, their performance according to their level of experience, and the CNN's area under the curve (AUC) of receiver operating characteristics (ROC).

RESULTS

The CNN revealed a sensitivity, specificity, and ROC AUC with corresponding 95% confidence intervals (CI) of 95.0% (95% CI 83.5% to 98.6%), 76.7% (95% CI 64.6% to 85.6%), and 0.918 (95% CI 0.866-0.970), respectively. In level I, the dermatologists' management decisions showed a mean sensitivity and specificity of 89.0% (95% CI 87.4% to 90.6%) and 80.7% (95% CI 78.8% to 82.6%). With level II information, the sensitivity significantly improved to 94.1% (95% CI 93.1% to 95.1%; P < 0.001), while the specificity remained unchanged at 80.4% (95% CI 78.4% to 82.4%; P = 0.97). When fixing the CNN's specificity at the mean specificity of the dermatologists' management decision in level II (80.4%), the CNN's sensitivity was almost equal to that of human raters, at 95% (95% CI 83.5% to 98.6%) versus 94.1% (95% CI 93.1% to 95.1%); P = 0.1. In contrast, dermatologists were outperformed by the CNN in their level I management decisions and level I and II diagnostic decisions. More experienced dermatologists frequently surpassed the CNN's performance.

CONCLUSIONS

Under less artificial conditions and in a broader spectrum of diagnoses, the CNN and most dermatologists performed on the same level. Dermatologists are trained to integrate information from a range of sources rendering comparative studies that are solely based on one single case image inadequate.

Skin lesions of face and scalp - Classification by a market-approved convolutional neural network in comparison with 64 dermatologists

BACKGROUND

The clinical differentiation of face and scalp lesions (FSLs) is challenging even for trained dermatologists. Studies comparing the diagnostic performance of a convolutional neural network (CNN) with dermatologists in FSL are lacking.

METHODS

A market-approved CNN (Moleanalyzer-Pro, FotoFinder Systems) was used for binary classifications of 100 dermoscopic images of FSL. The same lesions were used in a two-level reader study including 64 dermatologists (level I: dermoscopy only; level II: dermoscopy, clinical close-up images, textual information). Primary endpoints were the CNN's sensitivity and specificity in comparison with the dermatologists' management decisions in level II. Generalizability of the CNN results was tested by using four additional external data sets.

RESULTS

The CNN's sensitivity, specificity and ROC AUC were 96.2% [87.0%-98.9%], 68.8% [54.7%-80.1%] and 0.929 [0.880-0.978], respectively. In level II, the dermatologists' management decisions showed a mean sensitivity of 84.2% [82.2%-86.2%] and specificity of 69.4% [66.0%-72.8%]. When fixing the CNN's specificity at the dermatologists' mean specificity (69.4%), the CNN's sensitivity (96.2% [87.0%-98.9%]) was significantly higher than that of dermatologists (84.2% [82.2%-86.2%]; p < 0.001). Dermatologists of all training levels were outperformed by the CNN (all p < 0.001). In confirmation, the CNN's accuracy (83.0%) was significantly higher than dermatologists' accuracies in level II management decisions (all p < 0.001). The CNN's performance was largely confirmed in three additional external data sets but particularly showed a reduced specificity in one Australian data set including FSL on severely sun-damaged skin.

CONCLUSIONS

When applied as an assistant system, the CNN's higher sensitivity at an equivalent specificity may result in an improved early detection of face and scalp skin cancers.

Limitations of Deep Learning Attention Mechanisms in Clinical Research: Empirical Case Study Based on the Korean Diabetic Disease Setting

Background

Despite excellent prediction performance, noninterpretability has undermined the value of applying deep-learning algorithms in clinical practice. To overcome this limitation, attention mechanism has been introduced to clinical research as an explanatory modeling method. However, potential limitations of using this attractive method have not been clarified to clinical researchers. Furthermore, there has been a lack of introductory information explaining attention mechanisms to clinical researchers.

Objective

The aim of this study was to introduce the basic concepts and design approaches of attention mechanisms. In addition, we aimed to empirically assess the potential limitations of current attention mechanisms in terms of prediction and interpretability performance.

Methods

First, the basic concepts and several key considerations regarding attention mechanisms were identified. Second, four approaches to attention mechanisms were suggested according to a two-dimensional framework based on the degrees of freedom and uncertainty awareness. Third, the prediction performance, probability reliability, concentration of variable importance, consistency of attention results, and generalizability of attention results to conventional statistics were assessed in the diabetic classification modeling setting. Fourth, the potential limitations of attention mechanisms were considered.

Results

Prediction performance was very high for all models. Probability reliability was high in models with uncertainty awareness. Variable importance was concentrated in several variables when uncertainty awareness was not considered. The consistency of attention results was high when uncertainty awareness was considered. The generalizability of attention results to conventional statistics was poor regardless of the modeling approach.

Conclusions

The attention mechanism is an attractive technique with potential to be very promising in the future. However, it may not yet be desirable to rely on this method to assess variable importance in clinical settings. Therefore, along with theoretical studies enhancing attention mechanisms, more empirical studies investigating potential limitations should be encouraged.

Public Interest in Acne on the Internet: Comparison of Search Information From Google Trends and Naver

Background

Acne vulgaris is a common skin disease primarily affecting young adults. Given that the internet has become a major source of health information, especially among the young, the internet is a powerful tool of communication and has a significant influence on patients.

Objective

This study aimed to clarify the features of patients’ interest in and evaluate the quality of information about acne vulgaris on the internet.

Methods

We compared the search volumes on acne vulgaris with those of other dermatological diseases using Google Trends from January 2004 to August 2019. We also determined the search volumes for relevant keywords of acne vulgaris on Google and Naver and evaluated the quality of answers to the queries in KnowledgeiN.

Results

The regression analysis of Google Trends data demonstrated that the patients’ interest in acne vulgaris was higher than that for other dermatological diseases, such as atopic dermatitis (β=−20.33, 95% CI –22.27 to –18.39, P<.001) and urticaria (β=−27.09, 95% CI –29.03 to –25.15, P<.001) and has increased yearly (β=2.38, 95% CI 2.05 to 2.71, P<.001). The search volume for acne vulgaris was significantly higher in the summer than in the spring (β=–5.04, 95% CI –9.21 to –0.88, P=.018) and on weekends than on weekdays (β=–6.68, 95% CI –13.18 to –0.18, P=.044). The most frequently searched relevant keywords with “acne vulgaris” and “cause” were “stress,” “food,” and “cosmetics.” Among food, the 2 highest acne vulgaris–related keywords were milk and wheat in Naver and coffee and ramen in Google. The queries in Naver KnowledgeiN were mostly answered by a Korean traditional medicine doctor (53.4%) or the public (33.6%), but only 12.0% by dermatologists.

Conclusions

Physicians should be aware of patients’ interest in and beliefs about acne vulgaris to provide the best patient education and care, both online and in the clinic.

Prediction of the Age at Onset of Spinocerebellar Ataxia Type 3 with Machine Learning

BACKGROUND

In polyglutamine (polyQ) disease, the investigation of the prediction of a patient's age at onset (AAO) facilitates the development of disease-modifying intervention and underpins the delay of disease onset and progression. Few polyQ disease studies have evaluated AAO predicted by machine-learning algorithms and linear regression methods.

OBJECTIVE

The objective of this study was to develop a machine-learning model for AAO prediction in the largest spinocerebellar ataxia type 3/Machado-Joseph disease (SCA3/MJD) population from mainland China.

METHODS

In this observational study, we introduced an innovative approach by systematically comparing the performance of 7 machine-learning algorithms with linear regression to explore AAO prediction in SCA3/MJD using CAG expansions of 10 polyQ-related genes, sex, and parental origin.

RESULTS

Similar prediction performance of testing set and training set in each models were identified and few overfitting of training data was observed. Overall, the machine-learning-based XGBoost model exhibited the most favorable performance in AAO prediction over the traditional linear regression method and other 6 machine-learning algorithms for the training set and testing set. The optimal XGBoost model achieved mean absolute error, root mean square error, and median absolute error of 5.56, 7.13, 4.15 years, respectively, in testing set 1, with mean absolute error (4.78 years), root mean square error (6.31 years), and median absolute error (3.59 years) in testing set 2.

CONCLUSION

Machine-learning algorithms can be used to predict AAO in patients with SCA3/MJD. The optimal XGBoost algorithm can provide a good reference for the establishment and optimization of prediction models for SCA3/MJD or other polyQ diseases. © 2020 International Parkinson and Movement Disorder Society.

How Can Artificial Intelligence Make Medicine More Preemptive?

In this paper we propose the idea that Artificial intelligence (AI) is ushering in a new era of "Earlier Medicine," which is a predictive approach for disease prevention based on AI modeling and big data. The flourishing health care technological landscape is showing great potential-from diagnosis and prescription automation to the early detection of disease through efficient and cost-effective patient data screening tools that benefit from the predictive capabilities of AI. Monitoring the trajectories of both in- and outpatients has proven to be a task AI can perform to a reliable degree. Predictions can be a significant advantage to health care if they are accurate, prompt, and can be personalized and acted upon efficiently. This is where AI plays a crucial role in "Earlier Medicine" implementation.

Review of Machine Learning in Predicting Dermatological Outcomes

Artificial intelligence is a broad branch of computer science that has garnered significant interest in the field of medicine because of its problem solving, decision making and pattern recognition abilities. Machine learning, a subset of artificial intelligence, hones in on the ability of computers to receive data and learn for themselves, manipulating algorithms as they organize the information they are processing. Dermatology is at a particular advantage in the implementation of machine learning due to the availability of large clinical image databases that can be used for machine training and interpretation. While numerous studies have implemented machine learning in the diagnostic aspect of dermatology, less research has been conducted on the use of machine learning in predicting long-term outcomes in skin disease, with only a few studies published to date. Such an approach would assist physicians in selecting the best treatment methods, save patients' time, reduce treatment costs and improve the quality of treatment overall by reducing the amount of trial-and-error in the treatment process. In this review, we aim to provide a brief and relevant introduction to basic artificial intelligence processes, and to consolidate and examine the published literature on the use of machine learning in predicting clinical outcomes in dermatology.

Big Data in Ophthalmology

Big data is the fuel of mankind's fourth industrial revolution. Coupled with new technology such as artificial intelligence and deep learning, the potential of big data is poised to be harnessed to its maximal in years to come. In ophthalmology, given the data-intensive nature of this specialty, big data will similarly play an important role. Electronic medical records, administrative and health insurance databases, mega national biobanks, crowd source data from mobile applications and social media, and international epidemiology consortia are emerging forms of "big data" in ophthalmology. In this review, we discuss the characteristics of big data, its potential applications in ophthalmology, and the challenges in leveraging and using these data. Importantly, in the next phase of work, it will be pertinent to further translate "big data" findings into real-world applications, to improve quality of eye care, and cost-effectiveness and efficiency of health services in ophthalmology.

Machine Learning in Dermatology: Current Applications, Opportunities, and Limitations

Machine learning (ML) has the potential to improve the dermatologist's practice from diagnosis to personalized treatment. Recent advancements in access to large datasets (e.g., electronic medical records, image databases, omics), faster computing, and cheaper data storage have encouraged the development of ML algorithms with human-like intelligence in dermatology. This article is an overview of the basics of ML, current applications of ML, and potential limitations and considerations for further development of ML. We have identified five current areas of applications for ML in dermatology: (1) disease classification using clinical images; (2) disease classification using dermatopathology images; (3) assessment of skin diseases using mobile applications and personal monitoring devices; (4) facilitating large-scale epidemiology research; and (5) precision medicine. The purpose of this review is to provide a guide for dermatologists to help demystify the fundamentals of ML and its wide range of applications in order to better evaluate its potential opportunities and challenges.

LesionMap: A Method and Tool for the Semantic Annotation of Dermatological Lesions for Documentation and Machine Learning

Diagnosis and follow-up of patients in dermatology rely on visual cues. Documentation of skin lesions in dermatology is time-consuming and inaccurate. Digital photography is resource-intensive, difficult to standardize, and has privacy concerns. We propose a simple method—LesionMap—and an electronic health software tool—LesionMapper—for semantically annotating dermatological lesions on a body wireframe. We discuss how the type, distribution, and progression of lesions can be represented in a standardized way. The tool is an open-source JavaScript package that can be integrated into web-based electronic medical records. We believe that LesionMapper will facilitate documentation in dermatology that can be used for machine learning in a privacy-preserving manner.