Intelligence-based medicine最新文献

Background

In a recent comparative effectiveness trial, patients with chronic pain receiving cognitive behavioral therapy supported by artificial intelligence and digital feedback (AI-CBT-CP) were more likely to report clinically meaningful improvements in pain-related disability and intensity at six months than patients randomized to standard telephone CBT-CP. Concerns persist about the impact of AI and digital interventions among socially disadvantaged patients. We examined variation in the proportion of patients completing all treatment sessions and reporting clinically meaningful improvements in pain-related disability and intensity across subgroups of patients defined by social determinants of health (SDOH).

Methods

SDOH indicators included age, race, gender, education, income, marital status, geographic access, and clinical severity. Multivariate models with interaction terms tested SDOH indicators as potential moderators of treatment engagement and response to AI-CBT-CP versus standard telephone CBT-CP.

Findings

Roughly half of participants (52.9 %) were 65+ years of age, 10.8 % were women, and 19.1 % reported Black race or multiple racial identities. Relatively favorable session completion was observed among patients randomized to AI-CBT-CP across SDOH subgroup, with no groups more likely to complete all session weeks when receiving standard telephone CBT-CP. The relative benefits of AI-CBT-CP in terms of pain-related disability and intensity were generally confirmed across SDOH subgroups. AI-CBT-CP had a greater relative impact on pain-related disability among patients <65 years old (p = .002). In none of the SDOH subgroups, did standard telephone CBT-CP have a greater impact on pain-related disability or intensity than AI-CBT-CP.

Interpretation

These findings do not suggest that patients with SDOH disadvantages experience poorer treatment engagement or outcomes when offered CBT-CP supported by AI and digital feedback instead of standard telephone CBT-CP. AI-CBT-CP can help overcome treatment access barriers without exacerbating disparities, benefiting underserved populations with chronic pain.

Funding

US Department of Veterans Affairs Health Services Research and Development program.

Background and Objectives:

Methods:

Results:

Conclusion:

Immunotherapy is an important topic in healthcare as it affects patients' treatments for breast cancer, diabetes, and immunotherapy. However, immunotherapy for warts is less representative because of the lack of data. Machine learning is frequently utilised for treatment diagnosis by converting raw immunotherapy data into useful insights. Efficient classification of immunotherapy treatments is crucial for a productive diagnosis. This study considers immunotherapy with a data-driven and ’less is more perspective’. Despite using a portion of the available imbalance and complex data, the process of diagnosis of immunotherapy treatment is made reasonably precise by considering the parameters of accuracy, sensitivity, and specificity. The contribution of this study is focused on ”more is less” feature selection, which states that approximately 80 % of the effects or results of a system are caused by 20 % of the inputs. The features that contribute most to the classification of immunotherapy treatments are prioritised. This study proposes the implementation of Random Forest and Decision Trees for the classification of immunotherapy treatments. The relevant experimental medical data are explored as a case study. The experiments are conducted using Weka and Python data analysis tools, performing data preprocessing, class balancing, and feature selection. Random Forest performed better than the Decision Trees. By Applying Random Forest and utilising only one feature (time) as an input variable, a classification accuracy of 88.88 %, sensitivity of 95.45 %, and specificity of 60 % are attained. By using 12.5 % of the dataset, when implementing Random Forest together with ordinary feature selection, the diagnosis of immunotherapy treatments is become more efficient, despite using a portion of data features reasonable results are obtained.

Cancer patients can benefit from early detection and diagnosis. This study proposes a machine vision strategy for detecting breast cancer in ultrasound images and correcting several ultrasound difficulties such artifacts, speckle noise, and blurring. In quantitative evolution, edge preservation criteria were discovered to be superior to standard ones for hybrid anisotropic diffusion. A learnable super-resolution (SR) is inserted in the deep CNN to dig for extra possible information. The feature is fused with a pre-trained deep CNN model utilizing Gabor Wavelet Transform (GWT) and Local Binary Pattern (LBP). Machine learning (ML) techniques that are used to create these recommendation systems require well-balanced data in terms of class distribution, however most datasets in the real world are imbalanced. Imbalanced data forces a classifier to concentrate on the majority class while ignoring other classes of interest, lowering the predicted performance of any classification model. We propose a generative adversarial networks (GAN) strategy to overcome the data imbalance problem and improve the performance of recommendation systems in this research. Standard data is used to train this model, which assures a high level of resolution. In the testing phase, generalized data of varied resolution is used to evaluate the model's performance. It is discovered through cross-validation that a 5-fold method can successfully eliminate the overfitting problem. With an accuracy of 99.48 %, this suggested feature fusion technique indicates satisfactory performance when compared to current related works. Finally finding cancer region, researcher used U-Net architecture and extract cancer region from BC ultrasound images.

Background

Accurate assessment of trauma in the least time and efficient and effective treatment is gaining momentum in traumatology. Mapping the real-world practice patterns is essential in identifying and improving the quality of care for emergent time-dependent medical states like trauma.

Methods

The data mining solutions are extended to the National Trauma Registry of Iran (NTRI) event data by incorporating process mining techniques to ease the analysis, of the associations between clinical pathways and patient cohorts in understanding their performance. A total of 4498 cases, 44,344 events, and 104 different activities within the years 2017–2021 constitute the statistical data. Based on clinically relevant attributes and derived process characteristics the K-means clustering is applied to cohorts followed by comparing the clustering results and treatment pathways.

Results

The attributes influence treatment patterns in trauma care flows with the possibility of explaining the variations within cohorts' results. Although these attributes are not involved in the clustering algorithm, there exist meaningful correlations among the cohorts’ members in terms of type (final diagnostics) of injury, Injury Severity Score (minor: 1 < ISS<8; moderate: 9 < ISS<15; sever: 16 < ISS<24), Hospital Length of Stay (HLOS), and treatment activities.

Conclusion

Our findings provide more details on the existing process mining techniques and allow easy assessment of the quality of care at a given institution. This approach is an essential data analysis stage to improve complex care processes by proportioning the patient records into closely related groups applicable in target process-aware recommendation initiatives.

Objective

Benchmarking drug efficacy is a critical step in clinical trial design and planning. The challenge is that much of the data on efficacy endpoints is stored in scientific papers in free text form, so extraction of such data is currently a largely manual task. Our objective is to automate this task as much as possible.

Methods

In this study we have developed and optimised a framework to extract efficacy endpoints from text in scientific papers, using a machine learning approach.

Results

Our machine learning model predicts 25 classes associated with efficacy endpoints and leads to high F1 scores (harmonic mean of precision and recall) of 96.4 % on the test set, and 93.9 % and 93.7 % on two case studies.

Conclusion

These methods were evaluated against – and showed strong agreement with – subject matter experts and show significant promise in the future of automating the extraction of clinical endpoints from free text.

Significance

Clinical information extraction from text data is currently a laborious manual task which scales poorly and is prone to human error. Demonstrating the ability to extract efficacy endpoints automatically shows great promise for accelerating clinical trial design moving forwards.

Coronavirus disease 2019 (COVID-19) has become a pandemic all over the world and has spread rapidly. To distinguish between infected and non-infected areas, there is a critical need for segmentation methods that can identify infected areas from Chest Computed Tomography (CT) scans. In recent years, deep learning has become the most widely used approach for medical image segmentation, including the identification of infected areas in Chest CT scans. We propose an encoder-decoder based on the U-NET architecture for segmenting the MedSeg dataset, which contains lung CT scans. To study the effect of input dimensions on the model's output results, we gave CT images with dimensions of 224 × 224, 256 × 256, and 512 × 512 as inputs to the model. The results showed that 224 × 224 achieved higher results compared to 256 × 256 and 512 × 512, with a $dic{e}_{coef}$ of 81.36, accuracy of 87.65, sensitivity of 84.71, and specificity of 88.35. Additionally, the 224 × 224 input based on the proposed model achieved the highest number of correct answers compared to previous U-net methods. The proposed model can be applied as an effective screening tool to help primary service staff better refer suspected patients to specialists.

Diagnosing lung cancer is difficult due to the complexity of the nature of nodules. CT scan imaging is the most common imaging to diagnosis of lung cancer. Detection of nodules from these images is a challenge for radiologists and doctors. In recent years, neural networks have been developed for automatic, faster and more accurate diagnosis of diseases from medical images. In the present study, a new improved U-Net neural network is introduced for the automatic detection and segmentation of pulmonary nodules. The evaluation of this model has been done on LIDC-IDRI database. Our results have high values of recall, specificity and accuracy. The highest Recall value is 97.97 and is related to Juxtra-vascular. Specificity and accuracy for non-solid, partially solid and tiny has a value of 96.99.