Large Medical Datasets Research Articles

SUN-604 U.S. Prevalence & Mortality of Prader-Willi Syndrome: A Population-Based Study of Medical Claims

Prader-Willi syndrome (PWS) is a complex developmental genetic disorder associated with hypotonia, poor feeding in neonates, onset of hyperphagia in early childhood, and shorter overall life expectancy. Prior epidemiology studies of PWS have examined smaller populations, with limited research in a US population. The aim of this study was to provide a contemporary estimate of PWS prevalence and annual all-cause mortality in the US using a large administrative medical claims dataset. Methods: PWS patients were identified between 2012-2014 via the presence of ≥2 claims with a diagnosis code for PWS on medical claims provided by IQVIA™ Health Plan Claims Data and CMS Medicare fee-for-service claims. Patients were grouped into age bands including: 0-2, 3-8, 9-17, 18-26, 27-49, and ≥50. PWS prevalence and mortality rates were calculated for 2014, then 2018 US census data was used to project rates for the total US population. The presence of select diagnoses and procedures suggestive of a life-threatening event (e.g., mechanical ventilation) with a patient’s prompt disenrollment defined as death in the IQVIA data; vital status is indicated in Medicare data. Results: Overall US diagnosed PWS prevalence was 2.7 per 100k persons (or 1 per 37,037), a prevalence of 8,870 patients in the US in 2018. Diagnosed PWS prevalence 3.9, 5.2, 4.5, 4.2, 2.5, and 1.1 per 100k persons respectively for age bands 0-2, 3-8, 9-17, 18-26, 27-49, and ≥50. The median age of PWS patients was 21 years. The mortality rate was highest among diagnosed PWS patients aged 0-2 years and lowest among those aged 9-17 years and the overall mortality rate was 2.7%. For all respective age bands 0-2, 3-8, 9-17, 18-26, 27-49, and ≥50, the all-cause mortality rate was 5.4%, 3.0%, 1.4%, 2.1%, 2.4%, and 4.5%. The observed median age of death was 23 years (IQR 6-36) Conclusions: The diagnosed PWS prevalence of 1 per 37,037 persons estimated for the 2018 US population is comparable to the other reported US prevalence estimate. As the current study describes diagnosed patients, it likely represents a lower bound of true PWS prevalence. Annual PWS mortality is ≥3 times higher than the overall US population (2.7% vs 0.8%). This rate appears unchanged from mortality estimates reported for PWS populations in the last several decades despite significant advances in genetic testing and the availability of growth hormone therapies in the US. Aggressive management of serious comorbid conditions, especially in younger PWS patients, should be a clinical priority.

Splitting Large Medical Data Sets Based on Normal Distribution in Cloud Environment

The surge of medical and e-commerce applications has generated tremendous amount of data, which brings people to a so-called “Big Data” era. Different from traditional large data sets, the term “Big Data” not only means the large size of data volume but also indicates the high velocity of data generation. However, current data mining and analytical techniques are facing the challenge of dealing with large volume data in a short period of time. This paper explores the efficiency of utilizing the Normal Distribution (ND) method for splitting and processing large volume medical data in cloud environment, which can provide representative information in the split data sets. The ND-based new model consists of two stages. The first stage adopts the ND method for large data sets splitting and processing, which can reduce the volume of data sets. The second stage implements the ND-based model in a cloud computing infrastructure for allocating the split data sets. The experimental results show substantial efficiency gains of the proposed method over the conventional methods without splitting data into small partitions. The ND-based method can generate representative data sets, which can offer efficient solution for large data processing. The split data sets can be processed in parallel in Cloud computing environment.

Improving breast mass classification by shared data with domain transformation using a generative adversarial network

Training of a convolutional neural network (CNN) generally requires a large dataset. However, it is not easy to collect a large medical image dataset. The purpose of this study is to investigate the utility of synthetic images in training CNNs and to demonstrate the applicability of unrelated images by domain transformation. Mammograms showing 202 benign and 212 malignant masses were used for evaluation. To create synthetic data, a cycle generative adversarial network was trained with 599 lung nodules in computed tomography (CT) and 1430 breast masses on digitized mammograms (DDSM). A CNN was trained for classification between benign and malignant masses. The classification performance was compared between the networks trained with the original data, augmented data, synthetic data, DDSM images, and natural images (ImageNet dataset). The results were evaluated in terms of the classification accuracy and the area under the receiver operating characteristic curves (AUC). The classification accuracy improved from 65.7% to 67.1% with data augmentation. The use of an ImageNet pretrained model was useful (79.2%). Performance was slightly improved when synthetic images or the DDSM images only were used for pretraining (67.6 and 72.5%, respectively). When the ImageNet pretrained model was trained with the synthetic images, the classification performance slightly improved (81.4%), although the difference in AUCs was not statistically significant. The use of the synthetic images had an effect similar to the DDSM images. The results of the proposed study indicated that the synthetic data generated from unrelated lesions by domain transformation could be used to increase the training samples.

Cloud-computing and precision medicine: Big data offers big opportunities

Abstract Research on large shared medical datasets and data-driven research are gaining fast momentum and provide major opportunities for improving health systems as well as individual care. Such Open Data can shed light on the causes of disease and effects of treatment including adverse reactions side-effects of treatments, while also facilitating analyses tailored to an individual’s characteristics, known as personalized or precision medicine. Precision medicine treatments will take personalization to the next level and be effective for individual patients based on their genomic, environmental, and lifestyle factors. High-throughput sequencing technologies and open databases have made precision medicine possible but up to now only large research centres could handle the large scale technology needed for its processing. Aside from issues such as user trust, data privacy, transparency over the control of data ownership, and the implications of data analytics for personal privacy with potentially intrusive inferences, recent advances by Berkeley using open source Big Data technologies and Cloud Computing Services has allowed precision medicine studies to be conducted by small and agile research labs and researchers around the world.

Multimodal Medical Image Fusion Using Hybrid Layer Decomposition With CNN-Based Feature Mapping and Structural Clustering

Multimodal medical image fusion aims to integrate multisensory medical data into a single image for better prediction of the diagnostic details. This article presents a feature-level multimodal medical image fusion with the use of a two-scale ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> - ℓ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0</sub> hybrid layer decomposition scheme to maximize the structural details with the suppression of significant noise and artifacts. The proposed method uses a convolutional neural network (CNN) with consistency verification and structural patch clustering (fuzzy c-means-based) for the decomposed base and detail layer fusion, respectively. At first, a color space transform is used to separate the luminance and chrominance components extracted from both the source images and the luminance part of each image is decomposed. In the second step, a pretrained CNN model is used to extract the prominent features from each decomposed base layer component. For the output feature map, a regional energy-based activity measure is computed and generated a fusion score, which is refined in the consistency verification step to optimize the weight map for fusing the base layers. The two-scale detail layers are merged by the clustering-based prelearned dictionary for efficiently mapping of structural details of the layers. The color components associated with both the images are also combined using a color saliency measure. Finally, the fused base layer, detail layers, and color components are merged to get the resultant fused image. The experimental results justify the superiority of the proposed approach in terms of both subjective and objective assessments. The performance of the proposed scheme is tested on a large multimodal medical data set of MR-SPECT, MR-PET, and CT-MR neurological images. Furthermore, the proposed approach shows better results than the other state-of-the-art approaches with improved fusion quality and computational performances.

Using Data Analytics to Predict Hospital Mortality in Sepsis Patients

Predictive analytics can be used to anticipate the risks associated with some patients, and prediction models can be employed to alert physicians and allow timely proactive interventions. Recently, health care providers have been using different types of tools with prediction capabilities. Sepsis is one of the leading causes of in-hospital death in the United States and worldwide. In this study, the authors used a large medical dataset to develop and present a model that predicts in-hospital mortality among Sepsis patients. The predictive model was developed using a dataset of more than one million records of hospitalized patients. The independent predictors of in-hospital mortality were identified using the chi-square automatic interaction detector. The authors found that adding hospital attributes to the predictive model increased the accuracy from 82.08% to 85.3% and the area under the curve from 0.69 to 0.84, which is favorable compared to using only patients' attributes. The authors discuss the practical and research contributions of using a predictive model that incorporates both patient and hospital attributes in identifying high-risk patients.

Using Data Analytics to Predict Hospital Mortality in Sepsis Patients

Predictive analytics can be used to anticipate the risks associated with some patients, and prediction models can be employed to alert physicians and allow timely proactive interventions. Recently, health care providers have been using different types of tools with prediction capabilities. Sepsis is one of the leading causes of in-hospital death in the United States and worldwide. In this study, the authors used a large medical dataset to develop and present a model that predicts in-hospital mortality among Sepsis patients. The predictive model was developed using a dataset of more than one million records of hospitalized patients. The independent predictors of in-hospital mortality were identified using the chi-square automatic interaction detector. The authors found that adding hospital attributes to the predictive model increased the accuracy from 82.08% to 85.3% and the area under the curve from 0.69 to 0.84, which is favorable compared to using only patients' attributes. The authors discuss the practical and research contributions of using a predictive model that incorporates both patient and hospital attributes in identifying high-risk patients.

Automated semantic labeling of pediatric musculoskeletal radiographs using deep learning.

An automated method for identifying the anatomical region of an image independent of metadata labels could improve radiologist workflow (e.g., automated hanging protocols) and help facilitate the automated curation of large medical imaging data sets for machine learning purposes. Deep learning is a potential tool for this purpose. To develop and test the performance of deep convolutional neural networks (DCNN) for the automated classification of pediatric musculoskeletal radiographs by anatomical area. We utilized a database of 250 pediatric bone radiographs (50 each of the shoulder, elbow, hand, pelvis and knee) to train 5 DCNNs, one to detect each anatomical region amongst the others, based on ResNet-18 pretrained on ImageNet (transfer learning). For each DCNN, the radiographs were randomly split into training (64%), validation (12%) and test (24%) data sets. The training and validation data sets were augmented 30 times using standard preprocessing methods. We also tested our DCNNs on a separate test set of 100 radiographs from a single institution. Receiver operating characteristics (ROC) with area under the curve (AUC) were used to evaluate DCNN performances. All five DCNN trained for classification of the radiographs into anatomical region achieved ROC AUC of 1, respectively, for both test sets. Classification of the test radiographs occurred at a rate of 33 radiographs per s. DCNNs trained on a small set of images with 30 times augmentation through standard processing techniques are able to automatically classify pediatric musculoskeletal radiographs into anatomical region with near-perfect to perfect accuracy at superhuman speeds. This concept may apply to other body parts and radiographic views with the potential to create an all-encompassing semantic-labeling DCNN.

Applying deep matching networks to Chinese medical question answering: a study and a dataset

BackgroundMedical and clinical question answering (QA) is highly concerned by researchers recently. Though there are remarkable advances in this field, the development in Chinese medical domain is relatively backward. It can be attributed to the difficulty of Chinese text processing and the lack of large-scale datasets. To bridge the gap, this paper introduces a Chinese medical QA dataset and proposes effective methods for the task.MethodsWe first construct a large scale Chinese medical QA dataset. Then we leverage deep matching neural networks to capture semantic interaction between words in questions and answers. Considering that Chinese Word Segmentation (CWS) tools may fail to identify clinical terms, we design a module to merge the word segments and produce a new representation. It learns the common compositions of words or segments by using convolutional kernels and selects the strongest signals by windowed pooling.ResultsThe best performer among popular CWS tools on our dataset is found. In our experiments, deep matching models substantially outperform existing methods. Results also show that our proposed semantic clustered representation module improves the performance of models by up to 5.5% Precision at 1 and 4.9% Mean Average Precision.ConclusionsIn this paper, we introduce a large scale Chinese medical QA dataset and cast the task into a semantic matching problem. We also compare different CWS tools and input units. Among the two state-of-the-art deep matching neural networks, MatchPyramid performs better. Results also show the effectiveness of the proposed semantic clustered representation module.

Trends in checkpoint inhibitor therapy for advanced urothelial cell carcinoma (aUC) at the end of life: Insights from real-world practice.

395 Background: National guidelines recommend against chemotherapy use at the end of life. Five CPIs are now approved for aUC. We hypothesized that oncologists may have a lower threshold to initiate CPI in patients (pts) at the end of life, particularly among pts with poor performance status (PS), because of the perceived favorable toxicity profile of CPI. Methods: We conducted a secular trend analysis using the Flatiron Health Database, a large real-world electronic medical record-derived dataset. We used nonparametric tests of trend (p values reported as ptrend) to evaluate quarterly proportions of pts initiating CPI, chemotherapy, or no systemic therapy within 30 and 60 days of death, among 2959 pts diagnosed with aUC from 2015 to 2017. Results: 1637 pts died during follow-up and were eligible for analysis. 278 (17.0%) and 488 (29.8%) pts initiated a new line of systemic therapy in the last 30 and 60 days of life, respectively. The quarterly proportion of CPI initiators within 60 days of death increased from 1.0% to 23% during the study period ( ptrend &lt; 0.001), with corresponding decreases in chemotherapy initiation and no systemic therapy. After CPI approval in mid-2016, CPI initiation significantly increased among pts with Eastern Cooperative Oncology Group (ECOG) PS ≥ 2 ( ptrend = 0.02) but did not significantly change among pts with PS 0-1. Initiation of any systemic therapy at the end of life doubled (17.4% to 34.8%) over the study period, driven largely by end-of-life CPI use. Compared to chemotherapy initiators, CPI initiators had slightly worse PS (table). Conclusions: In patients with aUC, there has been a dramatic rise in CPI initiation at the end of life particularly among pts with poor PS. Existing guidelines on systemic therapy initiation near the end of life should account for the increasing use of CPI. [Table: see text]

Content based medical image retrieval using topic and location model.

Retrieval of medical images from an anatomically diverse dataset is a challenging task. Objective of our present study is to analyse the automated medical image retrieval system incorporating topic and location probabilities to enhance the performance. In this paper, we present an automated medical image retrieval system using Topic and Location Model. The topic information is generated using Guided Latent Dirichlet Allocation (GuidedLDA) method. A novel Location Model is proposed to incorporate the spatial information of visual words. We also introduce a new metric called position weighted Precision (wPrecision) to measure the rank order of the retrieved images. Experiments on two large medical image datasets - IRMA 2009 and Multimodal dataset - revealed that the proposed method outperforms existing medical image retrieval systems in terms of Precision and Mean Average Precision. The proposed method achieved better Mean Average Precision (86.74%) compared to the recent medical image retrieval systems using the Multimodal dataset with 7200 images. The proposed system achieved better Precision (97.5%) for top ten images compared to the recent medical image retrieval systems using IRMA 2009 dataset with 14,410 images. Supplementing spatial details of visual words to the Topic Model enhances the retrieval efficiency of medical images from large repositories. Such automated medical image retrieval systems can be used to assist physician to retrieve medical images with better precision compared to the state-of-the-art retrieval systems.

Medical data clustering based on particle swarm optimisation and genetic algorithm

Medical data clustering is popular scientific approach for finding hidden patterns from large medical dataset. Medical experts utilised these patterns to make clinical diagnosis for likelihood of a...

Content-based medical image retrieval by spatial matching of visual words

Content-Based Image Retrieval (CBIR) systems have recently emerged as one of the most promising and best image retrieval paradigms. To pacify the semantic gap associated with CBIR systems, the Bag of Visual Words (BoVW) techniques are now increasingly used. However, existing BoVW techniques fail to capture the location information of visual words effectively. This paper proposes an unsupervised Content-Based Medical Image Retrieval (CBMIR) framework based on the spatial matching of the visual words. The proposed method efficiently computes the spatial similarity of visual words using a novel similarity measure called the Skip Similarity Index. Experiments on three large medical datasets reveal promising results. The location-based correlation of visual words assists in more accurate and efficient retrieval of anatomically diverse and multimodal medical images than the state-of-the-art CBMIR systems.

1512O - Automated survival prediction in metastatic cancer patients using high-dimensional electronic medical record data

Background: Oncologists use patients’ life expectancy to guide medical decisions and may benefit from a tool that provides accurate, unbiased assessments of prognosis. Existing prognostic models generally use only a few predictor variables. We used a large electronic medical record dataset to train a prognostic model for patients with metastatic cancer. Methods: The model was trained and tested using data from 12,588 patients treated for metastatic cancer in the Stanford Health Care system from 2008-2017. Data sources included provider note text, labs, vital signs, procedures, medication orders, and diagnosis codes. Patients were divided randomly into training and test sets (80%/20% split). A regularized Cox proportional hazards model with 4,126 predictor variables was fit to the training set and evaluated on the test set. A landmarking approach was used due to the multiple observations per patient, with t0 set to the time of metastatic cancer diagnosis. Performance was also evaluated using 399 palliative radiation courses in test set patients. An existing published model that uses performance status, primary tumor site, and treated site was used as a baseline [Chow, JCO 2008:20;26(36)]. Results: From the first visit after metastatic cancer diagnosis, median follow-up was 14.5 months and median overall survival was 20.9 months. Patients were seen for 384,402 daily visits. The prognostic model’s C-index for overall survival was 0.79 in the test set (averaged across landmark times). For palliative radiation courses, the C-index was 0.75 (95% CI 0.72-0.78), compared to 0.64 (95% CI 0.60-0.67) for an existing published model (p < 0.001).Table: 1512OPredicted vs actual survival for 2,518 test set patients at landmark time t0 (first visit after diagnosis of metastatic cancer)Predicted median survival in monthsActual median survival in months (95% CI)0-3 (n = 106)1.3 (0.9-2.0)3.1-6 (n = 172)3.7 (2.5-5.2)6.1-12 (n = 382)6.7 (5.6-7.9)>12 (n = 1858)35.7 (31.8-39.1) Open table in a new tab Conclusions: The model showed high predictive performance, which was significantly better than that of an existing model. Because it is fully automated, the model can be used to examine providers’ practice patterns and could deployed in a decision support tool to help improve quality of care. Legal entity responsible for the study: Michael Gensheimer. Funding: National Institutes of Health. Disclosure: M.F. Gensheimer: Research funding: Varian Medical Systems, Philips Healthcare. E. Cho: Employee of Genentech. D. Rubin: Research funding: Philips Healthcare. D.T. Chang: Research funding, honoraria: Varian Medical Systems; Stock ownership: ViewRay. All other authors have declared no conflicts of interest.

Net ultrafiltration intensity and mortality in critically ill patients with fluid overload

BackgroundAlthough net ultrafiltration (UFNET) is frequently used for treatment of fluid overload in critically ill patients with acute kidney injury, the optimal intensity of UFNET is unclear. Among critically ill patients with fluid overload receiving renal replacement therapy (RRT), we examined the association between UFNET intensity and risk-adjusted 1-year mortality.MethodsWe selected patients with fluid overload ≥ 5% of body weight prior to initiation of RRT from a large academic medical center ICU dataset. UFNET intensity was calculated as the net volume of fluid ultrafiltered per day from initiation of either continuous or intermittent RRT until the end of ICU stay adjusted for patient hospital admission body weight. We stratified UFNET as low (≤ 20 ml/kg/day), moderate (> 20 to ≤ 25 ml/kg/day) or high (> 25 ml/kg/day) intensity. We adjusted for age, sex, body mass index, race, surgery, baseline estimated glomerular filtration rate, oliguria, first RRT modality, pre-RRT fluid balance, duration of RRT, time to RRT initiation from ICU admission, APACHE III score, mechanical ventilation use, suspected sepsis, mean arterial pressure on day 1 of RRT, cumulative fluid balance during RRT and cumulative vasopressor dose during RRT. We fitted logistic regression for 1-year mortality, Gray’s survival model and propensity matching to account for indication bias.ResultsOf 1075 patients, the distribution of high, moderate and low-intensity UFNET groups was 40.4%, 15.2% and 44.2% and 1-year mortality was 59.4% vs 60.2% vs 69.7%, respectively (p = 0.003). Using logistic regression, high-intensity compared with low-intensity UFNET was associated with lower mortality (adjusted odds ratio 0.61, 95% CI 0.41–0.93, p = 0.02). Using Gray’s model, high UFNET was associated with decreased mortality up to 39 days after ICU admission (adjusted hazard ratio range 0.50–0.73). After combining low and moderate-intensity UFNET groups (n = 258) and propensity matching with the high-intensity group (n = 258), UFNET intensity > 25 ml/kg/day compared with ≤ 25 ml/kg/day was associated with lower mortality (57% vs 67.8%, p = 0.01). Findings were robust to several sensitivity analyses.ConclusionsAmong critically ill patients with ≥ 5% fluid overload and receiving RRT, UFNET intensity > 25 ml/kg/day compared with ≤ 20 ml/kg/day was associated with lower 1-year risk-adjusted mortality. Whether tolerating intensive UFNET is just a marker for recovery or a mediator requires further research.

Supporting Diverse Challenges of Ageing with Digital Enhanced Living Solutions.

By the 2050, it is estimated that the proportion of people over the age of 80 will have risen from 3.9% to 9.1% of population of Organisation for Economic Cooperation and Development countries. A large proportion of these people will need significant help to manage various chronic illnesses, including dementia, heart disease, diabetes, limited physical movement and many others. Current approaches typically focus on acute episodes of illness and are not well designed to provide adequately for daily living care support. In our rapidly ageing society, a critical need exists for effective, affordable, scalable and safe in-home and in-residential care solutions leveraging a range of current and emerging sensor, interaction and integration technologies. Key aims are to support the ageing to live longer in their own homes; make daily challenges associated with ageing less limiting through use of technology supports; better support carers - both professional and family - in providing monitoring, proactive intervention, and community connectedness; enable in-home and in-residential care organisations to scale their support services and better use their workforces; and ultimately provide better quality of life. Deakin University researchers have been investigating a range of emerging technologies and platforms to realise this vision, which we in broad terms coin Digital Enhanced Living, in the ageing space but also supporting those with anxiety and depression, sleep disorders, various chronic diseases, recovery from injury, and various predictive analytics. A Smart Home solution, carried out in conjunction with a local start-up, has produced and trialled a novel sensor, interaction, and AI-based technology. Virtual Reality (VR) solutions have been used to support carers in the set-up of dementia-friendly homes, in conjunction with Alzheimers Australia. Activity and nutrition solutions, including the use of conversational agents, have been used to build dialogue to engage and change behaviour. Predictive analytics, in conjunction with major hospitals, have been applied to large medical datasets to better support professionals making judgements around discharge outcomes. A set of lessons have been learned from the design, deployment and trialing of these diverse solutions and new development approaches have been crafted to address the challenges faced. In particular, we found that there is a need to consider user emotional expectations as first-class citizens and create methodologies that consider the user needs during the creation of the software solutions. We find that quality and emotional aspects have to be engineered into the solution, rather than added after a technical solution is deployed.

A Touchless system for image visualization during surgery: preliminary experience in clinical settings.

Today clinicians may access large medical datasets, but very few systems have been designed to allow a practical and efficient exploration of data directly in critical medical environments such as operating rooms (OR). This work aims to assess during tests in laboratory and clinical settings a Surgery Touchless System (STS). This system allows clinicians to interact with medical images by using two different approaches: a gesture recognition and a voice recognition based system. These two methods are based on the use of a Microsoft Kinect and of a selective microphone, respectively. The STS allows navigating in a specifically designed interface, to perform several tasks, among others, to manipulate biomedical images. In this article, we assessed both the recognitions approaches in laboratory with 5 users. In addition, the STS was tested using only the voice-based recognition approach in clinical settings. The assessment was performed during three procedures by two interventionalradiologists. The five volunteers and the 2 radiologists filled two questionnaires to assess the system. The system usability was positively evaluated in laboratory tests. From clinical trials emerged that the STS was considered safe and useful by both the radiologists: they used the system an averaged number of times of 10 and 15 for patients, and found the system useful. These promising results allow considering this system useful for providing information not otherwise accessible and limiting the impact of human error during the operation. Future work will be focused on the use of the STS on a high number and different types of procedure.

Splitting Large Medical Data Sets based on Normal Distribution in Cloud Environment

Splitting Large Medical Data Sets based on Normal Distribution in Cloud Environment

Medical Data Clustering Based on Particle Swarm Optimization and Genetic Algorithm

Medical data clustering is popular scientific approach for finding hidden patterns from large medical dataset. Medical experts utilised these patterns to make clinical diagnosis for likelihood of a disease. Clustering groups the data objects of dataset into different groups based on data similarity within group is higher than other groups. In this work, a hybrid PSO-GA algorithm is developed for medical data clustering based on 'particle swarm optimisation (PSO) and genetic algorithm (GA)'. Hybrid PSO-GA performance has examined against K-means, PSO and GA with respect to six popular medical datasets namely iris, thyroid, breast cancer, heart, diabetes and pima adopted from UCI machine learning repository over three criterion namely, i.e., sum of intra cluster distance, error rate and CPU running time. Tabular and graphical results of simulation confirm hybrid PSO-GA technique for medical data clustering is superior against K-means, PSO and GA.

Adaptive Swarm Balancing Algorithms for rare-event prediction in imbalanced healthcare data.

Clinical data analysis and forecasting have made substantial contributions to disease control, prevention and detection. However, such data usually suffer from highly imbalanced samples in class distributions. In this paper, we aim to formulate effective methods to rebalance binary imbalanced dataset, where the positive samples take up only the minority. We investigate two different meta-heuristic algorithms, particle swarm optimization and bat algorithm, and apply them to empower the effects of synthetic minority over-sampling technique (SMOTE) for pre-processing the datasets. One approach is to process the full dataset as a whole. The other is to split up the dataset and adaptively process it one segment at a time. The experimental results reported in this paper reveal that the performance improvements obtained by the former methods are not scalable to larger data scales. The latter methods, which we call Adaptive Swarm Balancing Algorithms, lead to significant efficiency and effectiveness improvements on large datasets while the first method is invalid. We also find it more consistent with the practice of the typical large imbalanced medical datasets. We further use the meta-heuristic algorithms to optimize two key parameters of SMOTE. The proposed methods lead to more credible performances of the classifier, and shortening the run time compared to brute-force method.