Non-Invasive Cancer Detection Using Blood Test and Predictive Modeling Approach

Advances and Applications in Bioinformatics and Chemistry, Год журнала: 2025, Номер Volume 17, С. 159 - 178

Опубликована: Янв. 1, 2025

Purpose: The incidence of cancer, which is a serious public health concern, increasing. A predictive analysis driven by machine learning was integrated with haematology parameters to create method for the simultaneous diagnosis several malignancies at different stages. Patients and Methods: We analysed newly collected dataset from various hospitals in Jordan comprising 19,537 laboratory reports (6,280 cancer 13,257 noncancer cases). To clean obtain data ready modelling, preprocessing steps such as feature standardization missing value removal were used. Several cutting-edge classifiers employed prediction analysis. In addition, we experimented dataset's values using histogram gradient boosting (HGB) model. Results: ranking demonstrated ability distinguish patients healthy individuals based on hematological features WBCs, red blood cell (RBC) counts, platelet (PLT) addition age creatinine level. random forest (RF) classifier, followed linear discriminant (LDA) support vector (SVM), achieved highest accuracy (ranging 0.69 0.72 depending scenario investigated), reliably distinguishing between malignant benign conditions. HGB model showed improved performance dataset. Conclusion: After investigating number methods, an efficient screening platform non-invasive detection provided integration haematological indicators proper analytical data. Exploring deep methods future work, could provide insights into more complex patterns within dataset, potentially improving robustness predictions. Keywords: learning, complete count, RF model,

Язык: Английский

---

Deep Learning Approaches to Forecast Physical and Mental Deterioration During Chemotherapy in Patients with Cancer

Diagnostics, Год журнала: 2025, Номер 15(8), С. 956 - 956

Опубликована: Апрель 9, 2025

Background/Objectives: Predicting symptom escalation during chemotherapy is crucial for timely interventions and improved patient outcomes. This study employs deep learning models to predict the deterioration of 12 self-reported symptoms, categorized into physical (e.g., nausea, fatigue, pain) mental feeling blue, trouble thinking) groups. Methods: The analytical dataset comprises daily logs from individuals undergoing chemotherapy. To address class imbalance-where 84% cases showed no escalation-symptoms were grouped intervals 3 7 days. Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) trained on 80% data evaluated remaining 20%. Results: Results that 3-day yielded best predictive performance. CNNs excelled in predicting achieving 79.2% accuracy, 84.1% precision, 78.8% recall, an F1 score 81.4%. For GRU outperformed other models, with accuracy 77.2%, precision 71.6%, recall 62.2%, 66.6%. Performance declined longer due reduced temporal resolution fewer training samples, though remained relatively stable. Conclusions: findings emphasize advantage categorizing symptoms more tailored predictions demonstrate potential forecasting escalation. Integrating these clinical workflows could facilitate proactive management, allowing enhanced care

Язык: Английский

---

State of the “Art” in Precision Health Symptom Science Research

Seminars in Oncology Nursing, Год журнала: 2025, Номер unknown, С. 151906 - 151906

Опубликована: Май 1, 2025

Язык: Английский

---

AI-Driven Prediction of Symptom Trajectories in Cancer Care: A Deep Learning Approach for Chemotherapy Management

Bioengineering, Год журнала: 2024, Номер 11(11), С. 1172 - 1172

Опубликована: Ноя. 20, 2024

This study presents an advanced method for predicting symptom escalation in chemotherapy patients using Long Short-Term Memory (LSTM) networks and Convolutional Neural Networks (CNNs). The accurate prediction of is critical cancer care to enable timely interventions improve management enhance patients' quality life during treatment. analytical dataset consists daily self-reported logs from patients, including a wide range symptoms, such as nausea, fatigue, pain. original was highly imbalanced, with approximately 84% the data containing no escalation. were resampled into varying interval lengths address this imbalance model's ability detect (n = 3 n 7 days). allowed model predict significant changes severity across these intervals. results indicate that shorter intervals days) yielded highest overall performance, CNN achieving accuracy 81%, precision 87%, recall 80%, F1 score 83%. improvement over LSTM model, which had 79%, 85%, 82%. declined length increased, though remained relatively stable. findings demonstrate both CNN's temporospatial feature extraction LSTM's temporal modeling effectively capture patterns progression. By integrating predictive models digital health systems, healthcare providers can offer more personalized proactive care, enabling earlier may reduce burden treatment adherence. Ultimately, approach has potential significantly by providing real-time insights trajectories guiding clinical decision making.

Язык: Английский

---

Harnessing Natural Language Processing and High-Dimensional Clinical Notes to Detect Goals-of-Care and Surrogate-Designation Conversations

Clinical Nursing Research, Год журнала: 2024, Номер unknown

Опубликована: Окт. 31, 2024

Advance care planning, involving goals-of-care and surrogate-designation conversations, is crucial for patient-centered care. However, determining the optimal timing participants these conversations remains challenging. This study explored frequency, timing, predictors of documenting two advance planning elements, in clinical notes patients with advanced illness. In this retrospective observational study, we leveraged high-dimensional data natural language processing (NLP) to analyze predict presence or absence conversations. We included treated at a Midwestern United States hospital who had chronic conditions eventually passed away. manually labeled gold-standard dataset (

Язык: Английский

---

The Use of Machine Learning in Real-World Data: A Systematic Review of Disease Prediction and Management (Preprint)

Опубликована: Ноя. 17, 2024

Machine learning (ML) and big data analytics are revolutionizing healthcare, particularly in disease prediction, management, personalized care. With vast amounts of real-world (RWD) from sources like electronic health records (EHRs), patient registries, wearable devices, ML offers significant potential to improve clinical outcomes. However, quality, transparency, integration challenges remain. This study aims systematically review the use for prediction identifying most common methods, types, designs, evidence (RWE). A systematic followed PRISMA guidelines identify studies that utilized machine methods analyzing management. The focused on extracting related algorithms used, categories, types studies, RWE, such as devices. revealed frequently employed were Random Forest (RF), Logistic Regression (LR), Support Vector (SVM). These applied across various with cardiovascular diseases, cancers, neurological disorders being common. Real-world primarily originated EHRs, a predominant focus predictive modeling hold promise enhancing healthcare through better model interpretability, generalizability must be addressed integrate models fully into practice.

Язык: Английский

---

The Use of Machine Learning in Real-World Data: A Systematic Review of Disease Prediction and Management (Preprint)

JMIR Medical Informatics, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 17, 2024

Язык: Английский

---