Analyzing breast cancer invasive disease event classification through explainable artificial intelligence.
Ontology highlight
ABSTRACT: Introduction
Recently, accurate machine learning and deep learning approaches have been dedicated to the investigation of breast cancer invasive disease events (IDEs), such as recurrence, contralateral and second cancers. However, such approaches are poorly interpretable.Methods
Thus, we designed an Explainable Artificial Intelligence (XAI) framework to investigate IDEs within a cohort of 486 breast cancer patients enrolled at IRCCS Istituto Tumori "Giovanni Paolo II" in Bari, Italy. Using Shapley values, we determined the IDE driving features according to two periods, often adopted in clinical practice, of 5 and 10 years from the first tumor diagnosis.Results
Age, tumor diameter, surgery type, and multiplicity are predominant within the 5-year frame, while therapy-related features, including hormone, chemotherapy schemes and lymphovascular invasion, dominate the 10-year IDE prediction. Estrogen Receptor (ER), proliferation marker Ki67 and metastatic lymph nodes affect both frames.Discussion
Thus, our framework aims at shortening the distance between AI and clinical practice.
SUBMITTER: Massafra R
PROVIDER: S-EPMC9932275 | biostudies-literature | 2023
REPOSITORIES: biostudies-literature
Publications
Analyzing breast cancer invasive disease event classification through explainable artificial intelligence.
Massafra Raffaella R Fanizzi Annarita A Amoroso Nicola N Bove Samantha S Comes Maria Colomba MC Pomarico Domenico D Didonna Vittorio V Diotaiuti Sergio S Galati Luisa L Giotta Francesco F La Forgia Daniele D Latorre Agnese A Lombardi Angela A Nardone Annalisa A Pastena Maria Irene MI Ressa Cosmo Maurizio CM Rinaldi Lucia L Tamborra Pasquale P Zito Alfredo A Paradiso Angelo Virgilio AV Bellotti Roberto R Lorusso Vito V
Frontiers in medicine 20230202
<h4>Introduction</h4>Recently, accurate machine learning and deep learning approaches have been dedicated to the investigation of breast cancer invasive disease events (IDEs), such as recurrence, contralateral and second cancers. However, such approaches are poorly interpretable.<h4>Methods</h4>Thus, we designed an Explainable Artificial Intelligence (XAI) framework to investigate IDEs within a cohort of 486 breast cancer patients enrolled at IRCCS Istituto Tumori "Giovanni Paolo II" in Bari, It ...[more]
Similar Datasets
Project description:Electromyography (EMG) proves invaluable myoelectric manifestation in identifying neuromuscular alterations resulting from ischemic strokes, serving as a potential marker for diagnostics of gait impairments caused by ischemia. This study aims to develop an interpretable machine learning (ML) framework capable of distinguishing between the myoelectric patterns of stroke patients and those of healthy individuals through Explainable Artificial Intelligence (XAI) techniques. The research included 48 stroke patients (average age 70.6 years, 65% male) undergoing treatment at a rehabilitation center, alongside 75 healthy adults (average age 76.3 years, 32% male) as the control group. EMG signals were recorded from wearable devices positioned on the bicep femoris and lateral gastrocnemius muscles of both lower limbs during indoor ground walking in a gait laboratory. Boosting ML techniques were deployed to identify stroke-related gait impairments using EMG gait features. Furthermore, we employed XAI techniques, such as Shapley Additive Explanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), and Anchors to interpret the role of EMG variables in the stroke-prediction models. Among the ML models assessed, the GBoost model demonstrated the highest classification performance (AUROC: 0.94) during cross-validation with the training dataset, and it also overperformed (AUROC: 0.92, accuracy: 85.26%) when evaluated using the testing EMG dataset. Through SHAP and LIME analyses, the study identified that EMG spectral features contributing to distinguishing the stroke group from the control group were associated with the right bicep femoris and lateral gastrocnemius muscles. This interpretable EMG-based stroke prediction model holds promise as an objective tool for predicting post-stroke gait impairments. Its potential application could greatly assist in managing post-stroke rehabilitation by providing reliable EMG biomarkers and address potential gait impairment in individuals recovering from ischemic stroke.
| S-EPMC10935041 | biostudies-literature
Project description:Debilitating hearing loss (HL) affects ~6% of the human population. Only 20% of the people in need of a hearing assistive device will eventually seek and acquire one. The number of people that are satisfied with their Hearing Aids (HAids) and continue using them in the long term is even lower. Understanding the personal, behavioral, environmental, or other factors that correlate with the optimal HAid fitting and with users' experience of HAids is a significant step in improving patient satisfaction and quality of life, while reducing societal and financial burden. In SMART BEAR we are addressing this need by making use of the capacity of modern HAids to provide dynamic logging of their operation and by combining this information with a big amount of information about the medical, environmental, and social context of each HAid user. We are studying hearing rehabilitation through a 12-month continuous monitoring of HL patients, collecting data, such as participants' demographics, audiometric and medical data, their cognitive and mental status, their habits, and preferences, through a set of medical devices and wearables, as well as through face-to-face and remote clinical assessments and fitting/fine-tuning sessions. Descriptive, AI-based analysis and assessment of the relationships between heterogeneous data and HL-related parameters will help clinical researchers to better understand the overall health profiles of HL patients, and to identify patterns or relations that may be proven essential for future clinical trials. In addition, the future state and behavioral (e.g., HAids Satisfiability and HAids usage) of the patients will be predicted with time-dependent machine learning models to assist the clinical researchers to decide on the nature of the interventions. Explainable Artificial Intelligence (XAI) techniques will be leveraged to better understand the factors that play a significant role in the success of a hearing rehabilitation program, constructing patient profiles. This paper is a conceptual one aiming to describe the upcoming data collection process and proposed framework for providing a comprehensive profile for patients with HL in the context of EU-funded SMART BEAR project. Such patient profiles can be invaluable in HL treatment as they can help to identify the characteristics making patients more prone to drop out and stop using their HAids, using their HAids sufficiently long during the day, and being more satisfied by their HAids experience. They can also help decrease the number of needed remote sessions with their Audiologist for counseling, and/or HAids fine tuning, or the number of manual changes of HAids program (as indication of poor sound quality and bad adaptation of HAids configuration to patients' real needs and daily challenges), leading to reduced healthcare cost.
| S-EPMC9459083 | biostudies-literature
Project description:Background and purpose - Recent advances in artificial intelligence (deep learning) have shown remarkable performance in classifying non-medical images, and the technology is believed to be the next technological revolution. So far it has never been applied in an orthopedic setting, and in this study we sought to determine the feasibility of using deep learning for skeletal radiographs. Methods - We extracted 256,000 wrist, hand, and ankle radiographs from Danderyd's Hospital and identified 4 classes: fracture, laterality, body part, and exam view. We then selected 5 openly available deep learning networks that were adapted for these images. The most accurate network was benchmarked against a gold standard for fractures. We furthermore compared the network's performance with 2 senior orthopedic surgeons who reviewed images at the same resolution as the network. Results - All networks exhibited an accuracy of at least 90% when identifying laterality, body part, and exam view. The final accuracy for fractures was estimated at 83% for the best performing network. The network performed similarly to senior orthopedic surgeons when presented with images at the same resolution as the network. The 2 reviewer Cohen's kappa under these conditions was 0.76. Interpretation - This study supports the use for orthopedic radiographs of artificial intelligence, which can perform at a human level. While current implementation lacks important features that surgeons require, e.g. risk of dislocation, classifications, measurements, and combining multiple exam views, these problems have technical solutions that are waiting to be implemented for orthopedics.
| S-EPMC5694800 | biostudies-other
Project description:The distinction between Parkinson's disease (PD) and essential tremor (ET) tremors is subtle, posing challenges in differentiation. To accurately classify the PD and ET, BiLSTM-based recurrent neural networks are employed to classify between normal patients (N), PD patients, and ET patients using accelerometry data on their lower arm (L), hand (H), and upper arm (U) as inputs. The trained recurrent neural network (RNN) has reached 80% accuracy. The neural network is analyzed using layer-wise relevance propagation (LRP) to understand the internal workings of the neural network. A novel explainable AI method, called LRP-based approximate linear weights (ALW), is introduced to identify the similarities in relevance when assigning the class scores in the neural network. The ALW functions as a 2D kernel that linearly transforms the input data directly into the class scores, which significantly reduces the complexity of analyzing the neural network. This new classification method reconstructs the neural network's original function, achieving a 73% PD and ET tremor classification accuracy. By analyzing the ALWs, the correlation between each input and the class can also be determined. Then, the differentiating features can be subsequently identified. Since the input is preprocessed using short-time Fourier transform (STFT), the differences between the magnitude of tremor frequencies ranging from 3 to 30 Hz in the mean N, PD, and ET subjects are successfully identified. Aside from matching the current medical knowledge on frequency content in the tremors, the differentiating features also provide insights about frequency contents in the tremors in other frequency bands and body parts.
| S-EPMC10616175 | biostudies-literature
Project description:We demonstrate the capabilities of two model-agnostic local post-hoc model interpretability methods, namely breakDown (BD) and shapley (SHAP), to explain the predictions of a black-box classification learning model that establishes a quantitative relationship between chemical composition and multi-principal element alloys (MPEA) phase formation. We trained an ensemble of support vector machines using a dataset with 1,821 instances, 12 features with low pair-wise correlation, and seven phase labels. Feature contributions to the model prediction are computed by BD and SHAP for each composition. The resulting BD and SHAP transformed data are then used as inputs to identify similar composition groups using k-means clustering. Explanation-of-clusters by features reveal that the results from SHAP agree more closely with the literature. Visualization of compositions within a cluster using Ceteris-Paribus (CP) profile plots show the functional dependencies between the feature values and predicted response. Despite the differences between BD and SHAP in variable attribution, only minor changes were observed in the CP profile plots. Explanation-of-clusters by examples show that the clusters that share a common phase label contain similar compositions, which clarifies the similar-looking CP profile trends. Two plausible reasons are identified to describe this observation: (1) In the limits of a dataset with independent and non-interacting features, BD and SHAP show promise in recognizing MPEA composition clusters with similar phase labels. (2) There is more than one explanation for the MPEA phase formation rules with respect to the set of features considered in this work.
| S-EPMC9270422 | biostudies-literature
Project description:Abstract Aims The black box nature of artificial intelligence (AI) hinders the development of interpretable AI models that are applicable in clinical practice. We aimed to develop an AI model for classifying patients of reduced left ventricular ejection fraction (LVEF) from 12-lead electrocardiograms (ECG) with the decision-interpretability. Methods and results We acquired paired ECG and echocardiography datasets from the central and co-operative institutions. For the central institution dataset, a random forest model was trained to identify patients with reduced LVEF among 29 907 ECGs. Shapley additive explanations were applied to 7196 ECGs. To extract the model’s decision criteria, the calculated Shapley additive explanations values were clustered for 192 non-paced rhythm patients in which reduced LVEF was predicted. Although the extracted criteria were different for each cluster, these criteria generally comprised a combination of six ECG findings: negative T-wave inversion in I/V5–6 leads, low voltage in I/II/V4–6 leads, Q wave in V3–6 leads, ventricular activation time prolongation in I/V5–6 leads, S-wave prolongation in V2–3 leads, and corrected QT interval prolongation. Similarly, for the co-operative institution dataset, the extracted criteria comprised a combination of the same six ECG findings. Furthermore, the accuracy of seven cardiologists’ ECG readings improved significantly after watching a video explaining the interpretation of these criteria (before, 62.9% ± 3.9% vs. after, 73.9% ± 2.4%; P = 0.02). Conclusion We visually interpreted the model’s decision criteria to evaluate its validity, thereby developing a model that provided the decision-interpretability required for clinical application. Graphical Abstract Graphical Abstract
| S-EPMC10232279 | biostudies-literature
Project description:ObjectiveTo develop and evaluate a data-driven process to generate suggestions for improving alert criteria using explainable artificial intelligence (XAI) approaches.MethodsWe extracted data on alerts generated from January 1, 2019 to December 31, 2020, at Vanderbilt University Medical Center. We developed machine learning models to predict user responses to alerts. We applied XAI techniques to generate global explanations and local explanations. We evaluated the generated suggestions by comparing with alert's historical change logs and stakeholder interviews. Suggestions that either matched (or partially matched) changes already made to the alert or were considered clinically correct were classified as helpful.ResultsThe final dataset included 2 991 823 firings with 2689 features. Among the 5 machine learning models, the LightGBM model achieved the highest Area under the ROC Curve: 0.919 [0.918, 0.920]. We identified 96 helpful suggestions. A total of 278 807 firings (9.3%) could have been eliminated. Some of the suggestions also revealed workflow and education issues.ConclusionWe developed a data-driven process to generate suggestions for improving alert criteria using XAI techniques. Our approach could identify improvements regarding clinical decision support (CDS) that might be overlooked or delayed in manual reviews. It also unveils a secondary purpose for the XAI: to improve quality by discovering scenarios where CDS alerts are not accepted due to workflow, education, or staffing issues.
| S-EPMC10990514 | biostudies-literature
Project description:Understanding the pathological properties of dysregulated protein networks in individual patients' tumors is the basis for precision therapy. Functional experiments are commonly used, but cover only parts of the oncogenic signaling networks, whereas methods that reconstruct networks from omics data usually only predict average network features across tumors. Here, we show that the explainable AI method layer-wise relevance propagation (LRP) can infer protein interaction networks for individual patients from proteomic profiling data. LRP reconstructs average and individual interaction networks with an AUC of 0.99 and 0.93, respectively, and outperforms state-of-the-art network prediction methods for individual tumors. Using data from The Cancer Proteome Atlas, we identify known and potentially novel oncogenic network features, among which some are cancer-type specific and show only minor variation among patients, while others are present across certain tumor types but differ among individual patients. Our approach may therefore support predictive diagnostics in precision oncology by inferring "patient-level" oncogenic mechanisms.
| S-EPMC9174200 | biostudies-literature
Project description:State-of-the-art deep-learning systems use decision rules that are challenging for humans to model. Explainable AI (XAI) attempts to improve human understanding but rarely accounts for how people typically reason about unfamiliar agents. We propose explicitly modelling the human explainee via Bayesian teaching, which evaluates explanations by how much they shift explainees' inferences toward a desired goal. We assess Bayesian teaching in a binary image classification task across a variety of contexts. Absent intervention, participants predict that the AI's classifications will match their own, but explanations generated by Bayesian teaching improve their ability to predict the AI's judgements by moving them away from this prior belief. Bayesian teaching further allows each case to be broken down into sub-examples (here saliency maps). These sub-examples complement whole examples by improving error detection for familiar categories, whereas whole examples help predict correct AI judgements of unfamiliar cases.
| S-EPMC8110978 | biostudies-literature
Project description:AimCOVID-19 has revealed the need for fast and reliable methods to assist clinicians in diagnosing the disease. This article presents a model that applies explainable artificial intelligence (XAI) methods based on machine learning techniques on COVID-19 metagenomic next-generation sequencing (mNGS) samples.MethodsIn the data set used in the study, there are 15,979 gene expressions of 234 patients with COVID-19 negative 141 (60.3%) and COVID-19 positive 93 (39.7%). The least absolute shrinkage and selection operator (LASSO) method was applied to select genes associated with COVID-19. Support Vector Machine - Synthetic Minority Oversampling Technique (SVM-SMOTE) method was used to handle the class imbalance problem. Logistics regression (LR), SVM, random forest (RF), and extreme gradient boosting (XGBoost) methods were constructed to predict COVID-19. An explainable approach based on local interpretable model-agnostic explanations (LIME) and SHAPley Additive exPlanations (SHAP) methods was applied to determine COVID-19- associated biomarker candidate genes and improve the final model's interpretability.ResultsFor the diagnosis of COVID-19, the XGBoost (accuracy: 0.930) model outperformed the RF (accuracy: 0.912), SVM (accuracy: 0.877), and LR (accuracy: 0.912) models. As a result of the SHAP, the three most important genes associated with COVID-19 were IFI27, LGR6, and FAM83A. The results of LIME showed that especially the high level of IFI27 gene expression contributed to increasing the probability of positive class.ConclusionsThe proposed model (XGBoost) was able to predict COVID-19 successfully. The results show that machine learning combined with LIME and SHAP can explain the biomarker prediction for COVID-19 and provide clinicians with an intuitive understanding and interpretability of the impact of risk factors in the model.
| S-EPMC9889119 | biostudies-literature