Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Radiology Radiology

Artificial intelligence CAD tools in trauma imaging: a scoping review from the American Society of Emergency Radiology (ASER) AI/ML Expert Panel.

In Emergency radiology

BACKGROUND : AI/ML CAD tools can potentially improve outcomes in the high-stakes, high-volume model of trauma radiology. No prior scoping review has been undertaken to comprehensively assess tools in this subspecialty.

PURPOSE : To map the evolution and current state of trauma radiology CAD tools along key dimensions of technology readiness.

METHODS : Following a search of databases, abstract screening, and full-text document review, CAD tool maturity was charted using elements of data curation, performance validation, outcomes research, explainability, user acceptance, and funding patterns. Descriptive statistics were used to illustrate key trends.

RESULTS : A total of 4052 records were screened, and 233 full-text articles were selected for content analysis. Twenty-one papers described FDA-approved commercial tools, and 212 reported algorithm prototypes. Works ranged from foundational research to multi-reader multi-case trials with heterogeneous external data. Scalable convolutional neural network-based implementations increased steeply after 2016 and were used in all commercial products; however, options for explainability were narrow. Of FDA-approved tools, 9/10 performed detection tasks. Dataset sizes ranged from < 100 to > 500,000 patients, and commercialization coincided with public dataset availability. Cross-sectional torso datasets were uniformly small. Data curation methods with ground truth labeling by independent readers were uncommon. No papers assessed user acceptance, and no method included human-computer interaction. The USA and China had the highest research output and frequency of research funding.

CONCLUSIONS : Trauma imaging CAD tools are likely to improve patient care but are currently in an early stage of maturity, with few FDA-approved products for a limited number of uses. The scarcity of high-quality annotated data remains a major barrier.

Dreizin David, Staziaki Pedro V, Khatri Garvit D, Beckmann Nicholas M, Feng Zhaoyong, Liang Yuanyuan, Delproposto Zachary S, Klug Maximiliano, Spann J Stephen, Sarkar Nathan, Fu Yunting

2023-Mar-14

Artificial intelligence, Computer-aided detection, Emergency, Emergency radiology, Imaging, Machine learning, Radiology, Scoping review, Trauma

General General

Ensemble learning-based gene signature and risk model for predicting prognosis of triple-negative breast cancer.

In Functional & integrative genomics

Although medical science has been fully developed, due to the high heterogeneity of triple-negative breast cancer (TNBC), it is still difficult to use reasonable and precise treatment. In this study, based on local optimization-feature screening and genomics screening strategy, we screened 25 feature genes. In multiple machine learning algorithms, feature genes have excellent discriminative diagnostic performance among samples composed of multiple large datasets. After screening at the single-cell level, we identified genes expressed substantially in myeloid cells (MCGs) that have a potential association with TNBC. Based on MCGs, we distinguished two types of TNBC patients who showed considerable differences in survival status and immune-related characteristics. Immune-related gene risk scores (IRGRS) were established, and their validity was verified using validation cohorts. A total of 25 feature genes were obtained, among which CXCL9, CXCL10, CCL7, SPHK1, and TREM1 were identified as the result after single-cell level analysis and screening. According to these entries, the cohort was divided into MCA and MCB subtypes, and the two subtypes had significant differences in survival status and tumor-immune microenvironment. After Lasso-Cox screening, IDO1, GNLY, IRF1, CTLA4, and CXCR6 were selected for constructing IRGRS. There were significant differences in drug sensitivity and immunotherapy sensitivity among high-IRGRS and low-IRGRS groups. We revealed the dynamic relationship between TNBC and TIME, identified a potential biomarker called Granulysin (GNLY) related to immunity, and developed a multi-process machine learning package called "MPMLearning 1.0" in Python.

Li Tiancheng, Chen Siqi, Zhang Yuqi, Zhao Qianqian, Ma Kai, Jiang Xiwei, Xiang Rongwu, Zhai Fei, Ling Guixia

2023-Mar-14

Genomics, Immune, Machine learning, Myeloid cell, Triple-negative breast cancer

Radiology Radiology

Artificial intelligence in BreastScreen Norway: a retrospective analysis of a cancer-enriched sample including 1254 breast cancer cases.

In European radiology ; h5-index 62.0

OBJECTIVES : To compare results of selected performance measures in mammographic screening for an artificial intelligence (AI) system versus independent double reading by radiologists.

METHODS : In this retrospective study, we analyzed data from 949 screen-detected breast cancers, 305 interval cancers, and 13,646 negative examinations performed in BreastScreen Norway during the period from 2010 to 2018. An AI system scored the examinations from 1 to 10, based on the risk of malignancy. Results from the AI system were compared to screening results after independent double reading. AI score 10 was set as the threshold. The results were stratified by mammographic density.

RESULTS : A total of 92.7% of the screen-detected and 40.0% of the interval cancers had an AI score of 10. Among women with a negative screening outcome, 9.1% had an AI score of 10. For women with the highest breast density, the AI system scored 100% of the screen-detected cancers and 48.6% of the interval cancers with an AI score of 10, which resulted in a sensitivity of 80.9% for women with the highest breast density for the AI system, compared to 62.8% for independent double reading. For women with screen-detected cancers who had prior mammograms available, 41.9% had an AI score of 10 at the prior screening round.

CONCLUSIONS : The high proportion of cancers with an AI score of 10 indicates a promising performance of the AI system, particularly for women with dense breasts. Results on prior mammograms with AI score 10 illustrate the potential for earlier detection of breast cancers by using AI in screen-reading.

KEY POINTS : • The AI system scored 93% of the screen-detected cancers and 40% of the interval cancers with AI score 10. • The AI system scored all screen-detected cancers and almost 50% of interval cancers among women with the highest breast density with AI score 10. • About 40% of the screen-detected cancers had an AI score of 10 on the prior mammograms, indicating a potential for earlier detection by using AI in screen-reading.

Koch Henrik Wethe, Larsen Marthe, Bartsch Hauke, Kurz Kathinka Dæhli, Hofvind Solveig

2023-Mar-14

Artificial intelligence, Breast neoplasm, Mammographic density, Mammography, Mass screening

General General

Predicting Pain in People With Sickle Cell Disease in the Day Hospital Using the Commercial Wearable Apple Watch: Feasibility Study.

In JMIR formative research

BACKGROUND : Sickle cell disease (SCD) is a genetic red blood cell disorder associated with severe complications including chronic anemia, stroke, and vaso-occlusive crises (VOCs). VOCs are unpredictable, difficult to treat, and the leading cause of hospitalization. Recent efforts have focused on the use of mobile health technology to develop algorithms to predict pain in people with sickle cell disease. Combining the data collection abilities of a consumer wearable, such as the Apple Watch, and machine learning techniques may help us better understand the pain experience and find trends to predict pain from VOCs.

OBJECTIVE : The aim of this study is to (1) determine the feasibility of using the Apple Watch to predict the pain scores in people with sickle cell disease admitted to the Duke University SCD Day Hospital, referred to as the Day Hospital, and (2) build and evaluate machine learning algorithms to predict the pain scores of VOCs with the Apple Watch.

METHODS : Following approval of the institutional review board, patients with sickle cell disease, older than 18 years, and admitted to Day Hospital for a VOC between July 2021 and September 2021 were approached to participate in the study. Participants were provided with an Apple Watch Series 3, which is to be worn for the duration of their visit. Data collected from the Apple Watch included heart rate, heart rate variability (calculated), and calories. Pain scores and vital signs were collected from the electronic medical record. Data were analyzed using 3 different machine learning models: multinomial logistic regression, gradient boosting, and random forest, and 2 null models, to assess the accuracy of pain scores. The evaluation metrics considered were accuracy (F1-score), area under the receiving operating characteristic curve, and root-mean-square error (RMSE).

RESULTS : We enrolled 20 patients with sickle cell disease, all of whom identified as Black or African American and consisted of 12 (60%) females and 8 (40%) males. There were 14 individuals diagnosed with hemoglobin type SS (70%). The median age of the population was 35.5 (IQR 30-41) years. The median time each individual spent wearing the Apple Watch was 2 hours and 17 minutes and a total of 15,683 data points were collected across the population. All models outperformed the null models, and the best-performing model was the random forest model, which was able to predict the pain scores with an accuracy of 84.5%, and a RMSE of 0.84.

CONCLUSIONS : The strong performance of the model in all metrics validates feasibility and the ability to use data collected from a noninvasive device, the Apple Watch, to predict the pain scores during VOCs. It is a novel and feasible approach and presents a low-cost method that could benefit clinicians and individuals with sickle cell disease in the treatment of VOCs.

Stojancic Rebecca Sofia, Subramaniam Arvind, Vuong Caroline, Utkarsh Kumar, Golbasi Nuran, Fernandez Olivia, Shah Nirmish

2023-Mar-14

Apple Watch, consumer wearable, machine learning, mobile health, pain, predict, prediction, sickle cell disease, smartwatch, vaso-occlusive crises, wearable

General General

Enabling fully automated insulin delivery through meal detection and size estimation using Artificial Intelligence.

In NPJ digital medicine

We present a robust insulin delivery system that includes automated meal detection and carbohydrate content estimation using machine learning for meal insulin dosing called robust artificial pancreas (RAP). We conducted a randomized, single-center crossover trial to compare postprandial glucose control in the four hours following unannounced meals using a hybrid model predictive control (MPC) algorithm and the RAP system. The RAP system includes a neural network model to automatically detect meals and deliver a recommended meal insulin dose. The meal detection algorithm has a sensitivity of 83.3%, false discovery rate of 16.6%, and mean detection time of 25.9 minutes. While there is no significant difference in incremental area under the curve of glucose, RAP significantly reduces time above range (glucose >180 mg/dL) by 10.8% (P = 0.04) and trends toward increasing time in range (70-180 mg/dL) by 9.1% compared with MPC. Time below range (glucose <70 mg/dL) is not significantly different between RAP and MPC.

Mosquera-Lopez Clara, Wilson Leah M, El Youssef Joseph, Hilts Wade, Leitschuh Joseph, Branigan Deborah, Gabo Virginia, Eom Jae H, Castle Jessica R, Jacobs Peter G

2023-Mar-13

General General

Comprehensive analysis of clinical data for COVID-19 outcome estimation with machine learning models.

In Biomedical signal processing and control

COVID-19 is a global threat for the healthcare systems due to the rapid spread of the pathogen that causes it. In such situation, the clinicians must take important decisions, in an environment where medical resources can be insufficient. In this task, the computer-aided diagnosis systems can be very useful not only in the task of supporting the clinical decisions but also to perform relevant analyses, allowing them to understand better the disease and the factors that can identify the high risk patients. For those purposes, in this work, we use several machine learning algorithms to estimate the outcome of COVID-19 patients given their clinical information. Particularly, we perform 2 different studies: the first one estimates whether the patient is at low or at high risk of death whereas the second estimates if the patient needs hospitalization or not. The results of the analyses of this work show the most relevant features for each studied scenario, as well as the classification performance of the considered machine learning models. In particular, the XGBoost algorithm is able to estimate the need for hospitalization of a patient with an AUC-ROC of 0 . 8415 ± 0 . 0217 while it can also estimate the risk of death with an AUC-ROC of 0 . 7992 ± 0 . 0104 . Results have demonstrated the great potential of the proposal to determine those patients that need a greater amount of medical resources for being at a higher risk. This provides the healthcare services with a tool to better manage their resources.

Morís Daniel I, de Moura Joaquim, Marcos Pedro J, Rey Enrique Míguez, Novo Jorge, Ortega Marcos

2023-Jul

COVID-19, Classification, Clinical data, Feature selection, Machine learning