An interpretable schizophrenia diagnosis framework using machine learning and explainable artificial intelligence

Systems Science & Control Engineering, Год журнала: 2024, Номер 12(1)

Опубликована: Июнь 27, 2024

Schizophrenia is a complicated and multidimensional mental condition marked by wide range of emotional, cognitive, behavioural symptoms. Although the exact root cause schizophrenia unknown, experts believe that complex interaction genetic, environmental, neurobiological, neurodevelopmental, immune system dysfunctional elements are contributing factors. In healthcare, artificial intelligence (AI) used for analysing big datasets, enhance patient care, personalize treatment regimens, improve diagnostic accuracy, expedite administrative duties. Hence, ML has been to diagnose in this study. The term 'explainable intelligence' (XAI) describes development AI systems able provide understandable explanations their choices as well behaviours. our research paper, we harnessed power five diverse XAI methodologies: LIME (Local Interpretable Model-agnostic Explanations), SHAP (Shapley Additive exPlanations), ELI5 (Explain Like I'm 5), QLattice, Anchor. According (XAI), most significant attributes include age range, sex, presence triradius on left thumb, total number triradii, thenar region's palmar pattern. By enabling early intervention, automatic identification using can benefit patients, assisting doctors making precise diagnoses, medical personnel maximizing resource allocation care coordination.

Язык: Английский

---

Enhanced Data Mining and Visualization of Sensory-Graph-Modeled Datasets through Summarization

Sensors, Год журнала: 2024, Номер 24(14), С. 4554 - 4554

Опубликована: Июль 14, 2024

The acquisition, processing, mining, and visualization of sensory data for knowledge discovery decision support has recently been a popular area research exploration. Its usefulness is paramount because its relationship to the continuous involvement in improvement healthcare other related disciplines. As result this, huge amount have collected analyzed. These are made available community various shapes formats; their representation study form graphs or networks also an which many scholars focused on. However, large size such graph datasets poses challenges mining visualization. For example, from Bio–Mouse–Gene dataset, over 43 thousand nodes 14.5 million edges, non-trivial job. In this regard, summarizing provided useful alternative. Graph summarization aims provide efficient analysis complex large-sized data; hence, it beneficial approach. During summarization, all that similar structural properties merged together. doing so, traditional methods often overlook importance personalizing summary, would be helpful highlighting certain targeted nodes. Personalized context-specific scenarios require more tailored approach accurately capturing distinct patterns trends. Hence, concept personalized acquire concise depiction graph, emphasizing connections closer proximity specific set given target paper, we present faster algorithm (PGS) problem, named IPGS; designed facilitate enhanced effective domains, including biosensors. Our objective obtain compression ratio as one by state-of-the-art PGS algorithm, but manner. To achieve improve execution time current using weighted, locality-sensitive hashing, through experiments on eight publicly datasets. demonstrate effectiveness scalability IPGS while providing way, our contributes perspective summarization. We presented detailed was conducted investigate domain

Язык: Английский

---

Integrated ensemble CNN and explainable AI for COVID-19 diagnosis from CT scan and X-ray images

Scientific Reports, Год журнала: 2024, Номер 14(1)

Опубликована: Окт. 23, 2024

In light of the ongoing battle against COVID-19, while pandemic may eventually subside, sporadic cases still emerge, underscoring need for accurate detection from radiological images. However, limited explainability current deep learning models restricts clinician acceptance. To address this issue, our research integrates multiple CNN with explainable AI techniques, ensuring model interpretability before ensemble construction. Our approach enhances both accuracy and by evaluating advanced on largest publicly available X-ray dataset, COVIDx CXR-3, which includes 29,986 images, CT scan dataset SARS-CoV-2 Kaggle, a total 2,482 We also employed additional public datasets cross-dataset evaluation, thorough assessment performance across various imaging conditions. By leveraging methods including LIME, SHAP, Grad-CAM, Grad-CAM++, we provide transparent insights into decisions. model, DenseNet169, ResNet50, VGG16, demonstrates strong performance. For image sensitivity, specificity, accuracy, F1-score, AUC are recorded at 99.00%, 0.99, respectively. these metrics 96.18%, 0.9618, 0.96, methodology bridges gap between precision in clinical settings combining diversity explainability, promising enhanced disease diagnosis greater

Язык: Английский

---

Artificial Intelligence for Personalized Genetics and New Drug Development: Benefits and Cautions

Bioengineering, Год журнала: 2023, Номер 10(5), С. 613 - 613

Опубликована: Май 19, 2023

As the global health care system grapples with steadily rising costs, increasing numbers of admissions, and chronic defection doctors nurses from profession, appropriate measures need to be put in place reverse this course before it is too late [...].

Язык: Английский

---

Prediction of Urinary Tract Infection in IoT-Fog Environment for Smart Toilets Using Modified Attention-Based ANN and Machine Learning Algorithms

Applied Sciences, Год журнала: 2023, Номер 13(10), С. 5860 - 5860

Опубликована: Май 9, 2023

UTI (Urinary Tract Infection) has become common with maximum error rates in diagnosis. With the current progress on DM (Data Mining) based algorithms, several research projects have tried such algorithms due to their ability making optimal decisions and efficacy resolving complex issues. However, conventional failed attain accurate predictions improper feature selection. To resolve existing pitfalls, this intends employ suitable ML (Machine Learning)-based for predicting IoT-Fog environments, which will be applicable a smart toilet. Additionally, bio-inspired gained significant attention recent eras capability optimization Considering this, study proposes MFB-FA (Modified Flashing Behaviour-based Firefly Algorithm) This initializes FF (Firefly) population interchanges constant absorption coefficient value chaotic maps as chaos possesses an innate evade getting trapped local optima improvement determining global optimum. Further, GM (Gaussian Map) is taken into account moving all FFs optimum individual iteration. Due nature, algorithm better than other swarm intelligence approaches. Finally, classification undertaken by proposed MANN-AM Artificial Neural Network Attention Mechanism). The main intention proposing network involves its focus small data. Moreover, ANNs possess learning modelling non-linear relationships, present considers it. method compared internally using Random Forest, Naive Bayes K-Nearest Neighbour show of model. overall performance assessed regard standard metrics confirming prediction. model attained values accuracy 0.99, recall sensitivity 1, precision specificity 0.99 f1-score 0.99.

Язык: Английский

---

Predicting Multimorbidity Using Saudi Health Indicators (Sharik) Nationwide Data: Statistical and Machine Learning Approach

Healthcare, Год журнала: 2023, Номер 11(15), С. 2176 - 2176

Опубликована: Июль 31, 2023

The Saudi population is at high risk of multimorbidity. these morbidities can be reduced by identifying common modifiable behavioural factors. This study uses statistical and machine learning methods to predict factors for multimorbidity in the population. Data from 23,098 residents were extracted “Sharik” Health Indicators Surveillance System 2021. Participants asked about their demographics health indicators. Binary logistic models used determine predictors A backpropagation neural network model was further run using regression model. Accuracy measures checked training, validation, testing data. Females smokers had highest likelihood experiencing Age fruit consumption also played a significant role predicting Regarding accuracy, both algorithms yielded comparable outcomes. method (accuracy 80.7%) more accurate than (77%). Machine among adults, particularly Middle East region. Different later validated identified this study. These are helpful translated policymakers consider improvements public domain.

Язык: Английский

---

Breast Cancer Detection in the Equivocal Mammograms by AMAN Method

Applied Sciences, Год журнала: 2023, Номер 13(12), С. 7183 - 7183

Опубликована: Июнь 15, 2023

Breast cancer is a primary cause of human deaths among gynecological cancers around the globe. Though it can occur in both genders, far more common women. It disease which patient’s body cells breast start growing abnormally. has various kinds (e.g., invasive ductal carcinoma, lobular medullary, and mucinous), depend on turn into cancer. Traditional manual methods used to detect are not only time consuming but may also be expensive due shortage experts, especially developing countries. To contribute this concern, study proposed cost-effective efficient scheme called AMAN. based deep learning techniques diagnose its initial stages using X-ray mammograms. This system classifies two stages. In first stage, uses well-trained model (Xception) while extracting most crucial features from mammographs. The Xception pertained that well retrained by new data transfer approach. second involves gradient boost classify clinical specified set characteristics. Notably, experimental results satisfactory. attained an accuracy, area under curve (AUC), recall 87%, 95%, 86%, respectively, for mammography classification. For classification, achieved AUC 97% balanced accuracy 92%. Following these results, utilized relevant patients with high confidence.

Язык: Английский

---

Brixia Chest X-ray Score, Laboratory Parameters and Vaccination Status for Prediction of Mortality in COVID-19 Hospitalized Patients

Diagnostics, Год журнала: 2023, Номер 13(12), С. 2122 - 2122

Опубликована: Июнь 20, 2023

Chest X-ray has verified its role as a crucial tool in COVID-19 assessment due to practicability, especially emergency units, and Brixia score proven useful for pneumonia grading. The aim of our study was investigate correlations between main laboratory parameters, vaccination status, score, well confirm if is significant independent predictor unfavorable outcome (death) patients. designed cross-sectional multicentric study. It included patients with diagnosed infection who were hospitalized. This total 279 median age 62 years. only (adjusted odds ratio 1.148, p = 0.022). In addition, the results multiple linear regression analysis (R2 0.334, F 19.424, < 0.001) have shown that male gender (B 0.903, 0.046), severe 1.970, 0.001), lactate dehydrogenase 0.002, positive predictors, while albumin level -0.211, negative score. Our provide important information about factors influencing usefulness predicting These findings clinical relevance, epidemic circumstances.

Язык: Английский

---

Prediction of Health Risk Based on Multi-Level IOT Data Using Decision Trees

Опубликована: Сен. 14, 2023

This research paper presents a novel decision tree-based method for predicting health hazards based on multilevel Internet of Things (IoT). study's primary objective is to employ machine learning and deep techniques the field medical science in an effort make physicians' jobs easier have positive effect humanity. dataset consists 132 parameters from which 42 distinct disease types can be predicted. The data collected by (IoT) devices, are also used validation purposes. train tree classifier, then integrated into IoT-based device real-time risk prediction. Using classification metrics, accuracy model evaluated, feature importances analysed determine most significant risks. In addition, process selection employed eradicate less parameters, resulting refined model. multi-level IoT data, proposed demonstrates promising results with high hazards. contribute development intelligent healthcare systems facilitate early detection prevention.

Язык: Английский

---

Explainable AI in Healthcare: Systematic Review of Clinical Decision Support Systems

medRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Авг. 10, 2024

ABSTRACT This systematic review examines the evolution and current landscape of eXplainable Artificial Intelligence (XAI) in Clinical Decision Support Systems (CDSS), highlighting significant advancements identifying persistent challenges. Utilising PRISMA protocol, we searched major indexed databases such as Scopus, Web Science, PubMed, Cochrane Library, to analyse publications from January 2000 April 2024. timeframe captures progressive integration XAI CDSS, offering a historical technological overview. The covers datasets, application areas, machine learning models, explainable AI methods, evaluation strategies for multiple methods. Analysing 68 articles, uncover valuable insights into strengths limitations approaches, revealing research gaps providing actionable recommendations. We emphasise need more public advanced data treatment comprehensive evaluations interdisciplinary collaboration. Our findings stress importance balancing model performance with explainability enhancing usability tools medical practitioners. provides resource healthcare professionals, researchers, policymakers seeking develop evaluate effective, ethical decision-support systems clinical settings.

Язык: Английский

---