An Interpretable Three-Dimensional Artificial Intelligence Model for Computer-Aided Diagnosis of Lung Nodules in Computed Tomography Images.

Lung cancer is typically classified into small-cell carcinoma and non-small-cell carcinoma. Non-small-cell carcinoma accounts for approximately 85% of all lung cancers. Low-dose chest computed tomography (CT) can quickly and non-invasively diagnose lung cancer. In the era of deep learning, an artificial intelligence (AI) computer-aided diagnosis system can be developed for the automatic recognition of CT images of patients, creating a new form of intelligent medical service. For many years, lung cancer has been the leading cause of cancer-related deaths in Taiwan, with smoking and air pollution increasing the likelihood of developing the disease. The incidence of lung adenocarcinoma in never-smoking women has also increased significantly in recent years, resulting in an important public health problem. Early detection of lung cancer and prompt treatment can help reduce the mortality rate of patients with lung cancer. In this study, an improved 3D interpretable hierarchical semantic convolutional neural network named HSNet was developed and validated for the automatic diagnosis of lung cancer based on a collection of lung nodule images. The interpretable AI model proposed in this study, with different training strategies and adjustment of model parameters, such as cyclic learning rate and random weight averaging, demonstrated better diagnostic performance than the previous literature, with results of a four-fold cross-validation procedure showing calcification: 0.9873 ± 0.006, margin: 0.9207 ± 0.009, subtlety: 0.9026 ± 0.014, texture: 0.9685 ± 0.006, sphericity: 0.8652 ± 0.021, and malignancy: 0.9685 ± 0.006.

Artificial intelligence system for the detection of Barrett's esophagus.

BACKGROUND: Barrett's esophagus (BE), which has increased in prevalence worldwide, is a precursor for esophageal adenocarcinoma. Although there is a gap in the detection rates between endoscopic BE and histological BE in current research, we trained our artificial intelligence (AI) system with images of endoscopic BE and tested the system with images of histological BE. AIM: To assess whether an AI system can aid in the detection of BE in our setting. METHODS: Endoscopic narrow-band imaging (NBI) was collected from Chung Shan Medical University Hospital and Changhua Christian Hospital, resulting in 724 cases, with 86 patients having pathological results. Three senior endoscopists, who were instructing physicians of the Digestive Endoscopy Society of Taiwan, independently annotated the images in the development set to determine whether each image was classified as an endoscopic BE. The test set consisted of 160 endoscopic images of 86 cases with histological results. RESULTS: Six pre-trained models were compared, and EfficientNetV2B2 (accuracy [ACC]: 0.8) was selected as the backbone architecture for further evaluation due to better ACC results. In the final test, the AI system correctly identified 66 of 70 cases of BE and 85 of 90 cases without BE, resulting in an ACC of 94.37%. CONCLUSION: Our AI system, which was trained by NBI of endoscopic BE, can adequately predict endoscopic images of histological BE. The ACC, sensitivity, and specificity are 94.37%, 94.29%, and 94.44%, respectively.

An Improved Endoscopic Automatic Classification Model for Gastroesophageal Reflux Disease Using Deep Learning Integrated Machine Learning.

Gastroesophageal reflux disease (GERD) is a common digestive tract disease, and most physicians use the Los Angeles classification and diagnose the severity of the disease to provide appropriate treatment. With the advancement of artificial intelligence, deep learning models have been used successfully to help physicians with clinical diagnosis. This study combines deep learning and machine learning techniques and proposes a two-stage process for endoscopic classification in GERD, including transfer learning techniques applied to the target dataset to extract more precise image features and machine learning algorithms to build the best classification model. The experimental results demonstrate that the performance of the GerdNet-RF model proposed in this work is better than that of previous studies. Test accuracy can be improved from 78.8% ± 8.5% to 92.5% ± 2.1%. By enhancing the automated diagnostic capabilities of AI models, patient health care will be more assured.

Melanoma Detection Using XGB Classifier Combined with Feature Extraction and K-Means SMOTE Techniques.

Melanoma, a very severe form of skin cancer, spreads quickly and has a high mortality rate if not treated early. Recently, machine learning, deep learning, and other related technologies have been successfully applied to computer-aided diagnostic tasks of skin lesions. However, some issues in terms of image feature extraction and imbalanced data need to be addressed. Based on a method for manually annotating image features by dermatologists, we developed a melanoma detection model with four improvement strategies, including applying the transfer learning technique to automatically extract image features, adding gender and age metadata, using an oversampling technique for imbalanced data, and comparing machine learning algorithms. According to the experimental results, the improved strategies proposed in this study have statistically significant performance improvement effects. In particular, our proposed ensemble model can outperform previous related models.

Integrating Socioeconomic Status and Spatial Factors to Improve the Accessibility of Community Care Resources Using Maximum-Equity Optimization of Supply Capacity Allocation.

Health promotion empowers people, communities, and societies to take charge of their own health and quality of life. To strengthen community-based support, increase resource accessibility, and achieve the ideal of aging, this study targets the question of maximum equity with minimum values, taking distances and spatial and non-spatial factors into consideration. To compare disparities in the accessibility of community care resources and the optimization of allocation, methods for community care resource capacity were examined. This study also investigates units based on basic statistical area (BSA) to improve the limitation of larger reference locations (administrative districts) that cannot reflect the exact locations of people. The results show the capacity redistribution of each service point within the same total capacity, and the proposed method brings the population distribution of each demand to the best accessibility. Finally, the grading system of assessing accessibility scarcity allows the government to effectively categorize the prior improvement areas to achieve maximum equity under the same amount of care resources. There are 2046 (47.26%) and 396 (9.15%) BSAs that should be improved before and after optimization, respectively. Therefore, integrating socioeconomic status and spatial factors to assess accessibility of community-based care resources could provide comprehensive consideration for equal allocation.

A Deep Learning Model for Classification of Endoscopic Gastroesophageal Reflux Disease.

Gastroesophageal reflux disease (GERD) is a common disease with high prevalence, and its endoscopic severity can be evaluated using the Los Angeles classification (LA grade). This paper proposes a deep learning model (i.e., GERD-VGGNet) that employs convolutional neural networks for automatic classification and interpretation of routine GERD LA grade. The proposed model employs a data augmentation technique, a two-stage no-freezing fine-tuning policy, and an early stopping criterion. As a result, the proposed model exhibits high generalizability. A dataset of images from 464 patients was used for model training and validation. An additional 32 patients served as a test set to evaluate the accuracy of both the model and our trainees. Experimental results demonstrate that the best model for the development set exhibited an overall accuracy of 99.2% (grade A-B), 100% (grade C-D), and 100% (normal group) using narrow-band image (NBI) endoscopy. On the test set, the proposed model resulted in an accuracy of 87.9%, which was significantly higher than the results of the trainees (75.0% and 65.6%). The proposed GERD-VGGNet model can assist automatic classification of GERD in conventional and NBI environments and thereby increase the accuracy of interpretation of the results by inexperienced endoscopists.

The Role of Series Cholecystectomy in High Risk Acute Cholecystitis Patients Who Underwent Gallbladder Drainage.

Background: Cholecystectomy (CCY) is the only definitive therapy for acute cholecystitis. We conducted this study to evaluate which patients may not benefit from further CCY after percutaneous transhepatic gallbladder drainage (PTGBD) has been performed in acute cholecystitis patients. Methods: Acute cholecystitis patients with PTGBD treatment were selected from one million random samples from the National Health Insurance Research Database obtained between January 2004 and December 2010. Recurrent biliary events (RBEs), RBE-related medical costs, RBE-related mortality rate and an RBE-free survival curve were compared in patients who accepted CCY within 2 months and patients without CCY within 2 months after the index admission. Results: Three hundred and sixty-five acute cholecystitis patients underwent PTGBD at the index admission. A total of 190 patients underwent further CCY within 2 months after the index admission. The other 175 patients did not accept further CCY within 2 months after the index admission. RBE-free survival was significantly better in the CCY within 2 months group (60 vs. 42%, p < 0.001). The RBE-free survival of the CCY within 2 months group was similar to that of the no CCY within 2 months group in patients ≥ 80 years old and patients with a Charlson Comorbidity Index (CCI) score ≥ 9. Conclusions: We confirmed CCY after PTGBD reduced RBEs, RBE-related medical expenses, and the RBE-related mortality rate in patients with acute cholecystitis. In patients who accepted PTGBD, the RBE and survival benefits of subsequent CCY within 2 months became insignificant in patients ≥ 80 years old or with a CCI score ≥ 9.

Accessibility Assessment of Community Care Resources Using Maximum-Equity Optimization of Supply Capacity Allocation.

Equity in accessible healthcare is crucial for measuring health equity in community care policy. The most important objective of such a policy in Taiwan is empowering people and communities by improving health literacy and increasing access to healthcare resources. Using the nearest-neighbor two-step floating catchment area method, this study performed an accessibility assessment for community care resources before and after supply capacity optimization. For the target of maximum equity when allocating community care resources, taking maximum values, mean values and minimum values of the distances into consideration, three analytical allocation solutions for supply capability optimization were derived to further compare disparities in geographical accessibility. Three indicators, namely, the Gini coefficient, median minus mean and mean-squared error, were employed to assess the degree of optimization of geographical accessibility scores at the locations of the demand population and to determine the degree of geographic inequities in the allocation of community care resources. Our study proposed a method in which the minimum value of the distance is adopted as the approximate representation of distances between the service point and the locations of demand to determine the minimum value for supply capacity optimization. The study found that the method can effectively assess inequities in care resource allocation among urban and rural communities.

Investigating Health Equity and Healthcare Needs among Immigrant Women Using the Association Rule Mining Method.

Equitable access to healthcare services is a major concern among immigrant women. Thus, this study investigated the relationship between socioeconomic characteristics and healthcare needs among immigrant women in Taiwan. The secondary data was obtained from "Survey of Foreign and Chinese Spouses' Living Requirements, 2008," which was administered to 5848 immigrant women by the Ministry of the Interior, Taiwan. Additionally, descriptive statistics and significance tests were used to analyze the data, after which the association rule mining algorithm was applied to determine the relationship between socioeconomic characteristics and healthcare needs. According to the findings, the top three healthcare needs were providing medical allowances (52.53%), child health checkups (16.74%), and parental knowledge and pre- and post-natal guidance (8.31%). Based on the association analysis, the main barrier to the women's healthcare needs was "financial pressure." This study also found that nationality, socioeconomic status, and duration of residence were associated with such needs, while health inequality among aged immigrant women was due to economic and physical factors. Finally, the association analysis found that the women's healthcare problems included economic, socio-cultural, and gender weakness, while "economic inequality" and "women's health" were interrelated.

Symptomatic cholelithiasis patients have an increased risk of pancreatic cancer: A population-based study.

BACKGROUND AND AIM: Pancreatic cancer is a fatal disease; currently, the risk factor survey is not suitable for sporadic pancreatic cancer, which has neither family history nor the genetic analysis data. The aim of the present study was to evaluate the roles of cholelithiasis and cholelithiasis treatments on pancreatic cancer risk. METHODS: Symptomatic adult patients with an index admission of cholelithiasis were selected from one million random samples obtained between January 2005 and December 2009. The control group was matched with a 1:1 ratio for sex, age, chronic pancreatitis, and pancreatic cystic disease. Subsequent pancreatic cancer, which we defined as pancreatic cancer that occurred ≥ 6 months later, and total pancreatic cancer events were calculated in the cholelithiasis and control groups. The cholelithiasis group was further divided into endoscopic sphincterotomy/endoscopic papillary balloon dilatation, cholecystectomy, endoscopic sphincterotomy/endoscopic papillary balloon dilatation and cholecystectomy, and no-intervention groups for evaluation. RESULTS: The cholelithiasis group and the matched control group included 8265 adults. The cholelithiasis group contained 86 cases of diagnosed pancreatic cancer, and the control group contained 8 cases (P < 0.001). The incidence rate ratio (IRR) of subsequent pancreatic cancer was significantly higher in the cholelithiasis group than in the control group (IRR: 5.28, P < 0.001). The IRR of subsequent pancreatic cancer was higher in the no-intervention group comparing with cholecystectomy group (IRR = 3.21, P = 0.039) but was similar in other management subgroups. CONCLUSION: Symptomatic cholelithiasis is a risk factor for pancreatic cancer; the risk is similar regardless of the intervention chosen for cholelithiasis.

Cholecystectomy reduces subsequent cholangiocarcinoma risk in choledocholithiasis patients undergoing endoscopic intervention.

BACKGROUND: Cholangiocarcinoma is a disease with a high mortality rate. Our previous study revealed that cholelithiasis patients who undergo endoscopic sphincterotomy (ES)/endoscopic papillary balloon dilatation are at a higher risk for subsequent cholangiocarcinoma than cholelithiasis patients who undergo cholecystectomy. AIM: To clarify the relationship between recurrent biliary events and subsequent cholangiocarcinoma risk in choledocholithiasis patients. METHODS: From one million random cases in the Taiwan National Health Insurance Research Database 2004-2011, we selected symptomatic choledocholithiasis patients older than 18 years who were admitted from January 2005 to December 2009 (study group). Cases for a control group were defined as individuals who had never been diagnosed with cholelithiasis, matched by sex and age in a 1:3 ratio. The study group was further divided into ES/endoscopic papillary balloon dilatation, both ES/endoscopic papillary balloon dilatation and cholecystectomy, and no intervention groups. RESULTS: We included 2096 choledocholithiasis patients without previous intervention or cholangiocarcinoma. A total of 12 (2.35%), 11 (0.74%), and 1 (1.00%) subsequent cholangiocarcinoma cases were diagnosed among 511 ES/endoscopic papillary balloon dilatation patients, 1485 patients with no intervention, and 100 ES/endoscopic papillary balloon dilatation and cholecystectomy patients, respectively. The incidence rates of recurrent biliary event were 527.79/1000 person-years and 286.69/1000 person-years in the subsequent cholangiocarcinoma and no cholangiocarcinoma group, showing a high correlation between subsequent cholangiocarcinoma risk and recurrent biliary events. CONCLUSION: Choledocholithiasis patients who undergo further cholecystectomy after ES/endoscopic papillary balloon dilatation have decreased subsequent cholangiocarcinoma risk due to reduced recurrent biliary events.

Assessment on Distributional Fairness of Physical Rehabilitation Resource Allocation: Geographic Accessibility Analysis Integrating Google Rating Mechanism.

Identifying and treating co-existing diseases are essential in healthcare for the elderly, while physical rehabilitation care teams can provide interdisciplinary geriatric care for the elderly. To evaluate the appropriateness of demand and supply between the population at demand and physical rehabilitation resources, a comparative analysis was carried out in this study. Our study applied seven statistical indices to assess five proposed methods those considered different factors for geographic accessibility analysis. Google ratings were included in the study as a crucial factor of choice probability in the equation for calculating the geographic accessibility scores, because people's behavioral decisions are increasingly dependent on online rating information. The results showed that methods considering distances, the capacity of hospitals, and Google ratings' integrally generated scores, are in better accordance with people's decision-making behavior when they determine which resources of physical rehabilitation to use. It implies that concurrent considerations of non-spatial factors (online ratings and sizes of resource) are important. Our study proposed an integrated assessment method of geographical accessibility scores, which includes the spatial distribution, capacity of resources and online ratings in the mechanism. This research caters to countries that provide citizens with a higher degree of freedom in their medical choices and allows these countries to improve the fairness of resource allocation, raise the geographic accessibilities of physical rehabilitation resources, and promote aging in place.

Comorbidities and Outcome of Alcoholic and Non-Alcoholic Liver Cirrhosis in Taiwan: A Population-Based Study.

The prognosis of different etiologies of liver cirrhosis (LC) is not well understood. Previous studies performed on alcoholic LC-dominated cohorts have demonstrated a few conflicting results. We aimed to compare the outcome and the effect of comorbidities on survival between alcoholic and non-alcoholic LC in a viral hepatitis-dominated LC cohort. We identified newly diagnosed alcoholic and non-alcoholic LC patients, aged ≥40 years old, between 2006 and 2011, by using the Longitudinal Health Insurance Database. The hazard ratios (HRs) were calculated using the Cox proportional hazards model and the Kaplan-Meier method. A total of 472 alcoholic LC and 4313 non-alcoholic LC patients were identified in our study cohort. We found that alcoholic LC patients were predominantly male (94.7% of alcoholic LC and 62.6% of non-alcoholic LC patients were male) and younger (78.8% of alcoholic LC and 37.4% of non-alcoholic LC patients were less than 60 years old) compared with non-alcoholic LC patients. Non-alcoholic LC patients had a higher rate of concomitant comorbidities than alcoholic LC patients (79.6% vs. 68.6%, p < 0.001). LC patients with chronic kidney disease demonstrated the highest adjusted HRs of 2.762 in alcoholic LC and 1.751 in non-alcoholic LC (all p < 0.001). In contrast, LC patients with hypertension and hyperlipidemia had a decreased risk of mortality. The six-year survival rates showed no difference between both study groups (p = 0.312). In conclusion, alcoholic LC patients were younger and had lower rates of concomitant comorbidities compared with non-alcoholic LC patients. However, all-cause mortality was not different between alcoholic and non-alcoholic LC patients.

Risk of cholangiocarcinoma in patients undergoing therapeutic endoscopic retrograde cholangiopancreatography or cholecystectomy: A population based study.

BACKGROUND: Cholangiocarcinoma is a highly lethal disease that had been underestimated in the past two decades. Many risk factors are well documented for in cholangiocarcinoma, but the impacts of advanced biliary interventions, like endoscopic sphincterotomy (ES), endoscopic papillary balloon dilatation (EPBD), and cholecystectomy, are inconsistent in the previous literature. AIM: To clarify the risks of cholangiocarcinoma after ES/EPBD, cholecystectomy or no intervention for cholelithiasis using the National Health Insurance Research Database (NHIRD). METHODS: From data of NHIRD 2004-2011 in Taiwan, we selected 7938 cholelithiasis cases as well as 23814 control group cases (matched by sex and age in a 1:3 ratio). We compared the previous risk factors of cholangiocarcinoma and cholangiocarcinoma rate in the cholelithiasis and control groups. The incidences of total and subsequent cholangiocarcinoma were calculated in ES/EPBD patients, cholecystectomy patients, cholelithiasis patients without intervention, and groups from the normal population. RESULTS: In total, 537 cases underwent ES/EPBD, 1743 cases underwent cholecystectomy, and 5658 cholelithiasis cases had no intervention. Eleven (2.05%), 37 (0.65%), and 7 (0.40%) subsequent cholangiocarcinoma cases were diagnosed in the ES/EPBD, no intervention, and cholecystectomy groups, respectively, and the odds ratio for subsequent cholangiocarcinoma was 3.13 in the ES/EPBD group and 0.61 in the cholecystectomy group when compared with the no intervention group. CONCLUSION: In conclusion, symptomatic cholelithiasis patients who undergo cholecystectomy can reduce the incidence of subsequent cholangiocarcinoma, while cholelithiasis patients who undergo ES/EPBD are at a great risk of subsequent cholangiocarcinoma according to our findings.

Role of Cholecystectomy in Choledocholithiasis Patients Underwent Endoscopic Retrograde Cholangiopancreatography.

There are no clinical guidelines for the timing of cholecystectomy (CCY) after performing therapeutic endoscopic retrograde cholangiopancreatography (ERCP) for choledocholithiasis. We tried to analyze the clinical practice patterns, medical expenses, and subsequent outcomes between the early CCY, delayed CCY, and no CCY groups of patients. 1827 choledocholithiasis patients who underwent therapeutic ERCP were selected from the nationwide population databases of two million random samples. These patients were further divided into early CCY, delayed CCY, and no CCY performed. In our analysis, 1440 (78.8%) of the 1827 patients did not undergo CCY within 60 days of therapeutic ERCP, and only 239 (13.1%) patients underwent CCY during their index admission. The proportion of laparoscopic CCY increased from 37.2% to 73.6% in the delayed CCY group. There were no significant differences (p = 0.934) between recurrent biliary event (RBE) rates with or without early CCY within 60 days of ERCP. RBE event-free survival rates were significantly different in the early CCY (85.04%), delayed CCY (89.54%), and no CCY (64.45%) groups within 360 days of ERCP. The method of delayed CCY can reduce subsequent RBEs and increase the proportion of laparoscopic CCY with similar medical expenses to early CCY in Taiwan's general practice environment.

Major bleeding risk of endoscopic sphincterotomy versus endoscopic papillary balloon dilatation in hemodialysis patients.

BACKGROUND/AIMS: Endoscopic sphincterotomy (EST) and endoscopic papillary balloon dilatation (EPBD) are used for therapeutic endoscopic retrograde cholangiopancreatography (ERCP). The postprocedure bleeding rate for EPBD is low in the normal population; however, this bleeding rate in a group of patients prone to bleeding, such as patients with end-stage renal disease, is not well-established. We therefore evaluated the post-EST and post-EPBD bleeding rate among hemodialysis (HD) patients based on data from Taiwan's National Health Insurance Research Database (NHIRD). PATIENTS AND METHODS: The NHIRD entries for a population of 2 million were screened for patients who had a catastrophic illness card for HD between 1st January 2004 and 31st December 2011 and these patients were enrolled as research subjects. The rates of major gastrointestinal tract bleeding events appearing within 14 days after EST or EPBD were compared between HD and non-HD patients. RESULTS: A total of 3561 patients, over 18 years of age and without liver cirrhosis or hematologic diseases, underwent 3826 EST and 280 EPBD procedures during the 8 calendar years selected for our analysis. The total post-ERCP major bleeding rate was much higher in HD than in non-HD patients (8.64% vs. 2.16%, P < 0.0001). The rate of postprocedure major bleeding events was lower for non-HD patients who underwent EPBD than those who underwent EST (0.75% vs. 2.26%; P = 0.049), whereas the postprocedure major bleeding event rates were similar in HD patients who underwent either EPBD or EST (8.70% vs. 8.33%; P = 0.484). CONCLUSION: Post-ERCP, post-EST, and post-EPBD major bleeding rates were all higher in HD patients in this study. EPBD resulted in lower postprocedure major bleeding events than EST in the non-HD population, but it failed to provide the reduction in bleeding events needed to perform endoscopic hemostasis in HD patients.

Favorable clinical outcome of nonalcoholic liver cirrhosis patients with coronary artery disease: A population-based study.

AIM: To elucidate the prevalence and risk of mortality of nonalcoholic liver cirrhosis (LC) patients with coronary artery disease (CAD). METHODS: The study cohort included newly diagnosed nonalcoholic LC patients age ≥ 40 years old without a diagnosis of CAD from 2006 until 2011 from a longitudinal health insurance database. The mean follow-up period for the study cohort was 1152 ± 633 d. The control cohort was matched by sex, age, residence, and index date. Hazard ratios (HRs) were calculated using the Cox proportional hazard model and the Kaplan-Meier method. RESULTS: After exclusion, a total of 3409 newly diagnosed nonalcoholic cirrhotic patients were identified from one million samples from the health insurance database. We found that CAD (5.1% vs 17.4%) and hyperlipidemia (20.6% vs 24.1%) were less prevalent in nonalcoholic LC patients than in normal subjects (all P < 0.001), whereas other comorbidities exhibited an increased prevalence. Among the comorbidities, chronic kidney disease exhibited the highest risk for mortality (adjusted HR (AHR) = 1.76; 95%CI: 1.55-2.00, P < 0.001). Ascites or peritonitis exhibited the highest risk of mortality among nonalcoholic cirrhotic patients (AHR = 2.34; 95%CI: 2.06-2.65, P < 0.001). Finally, a total of 170 patients developed CAD after a diagnosis of nonalcoholic LC. The AHR of CAD in nonalcoholic LC patients was 0.56 (95%CI: 0.43-0.74, P < 0.001). The six-year survival rates for nonalcoholic LC patients with and without CAD were 52% and 50%, respectively (P = 0.012). CONCLUSION: We conclude that CAD was less prevalent and associated with a reduced risk of mortality in nonalcoholic cirrhotic patients.

Evaluating Disparities in Elderly Community Care Resources: Using a Geographic Accessibility and Inequality Index.

This study evaluated geographic accessibility and utilized assessment indices to investigate disparities in elderly community care resource distribution. The data were derived from Taiwanese governmental data in 2017, including 3,148,283 elderly individuals (age 65+), 7681 villages, and 1941 community care centers. To identify disparities in geographic accessibility, we compared the efficacy of six measurements and proposed a composite index to identify levels of resource inequality from the Gini coefficient and “median-mean” skewness. Low village-level correlation (0.038) indicated inconsistencies between the demand populations and community care center distribution. Method M6 (calculated accessibility of nearest distance-decay accounting for population of villages, supplier loading, and elderly walkability) was identified as the most comprehensive disparity measurement. Community care policy assessment requires a comprehensive and weighted calculation process, including the elderly walkability distance-decay factor, demand population, and supplier loading. Three steps were suggested for elderly policy planning and improvement in future.

Decision tree based diagnostic system for moderate to severe obstructive sleep apnea.

One of the major modern medical issues, obstructive sleep apnea (OSA), particularly at moderate to severe levels, may potentially cause cardiovascular morbidity and mortality. However, polysomnography (PSG), a gold standard tool in diagnosing OSA, is cumbersome, has limited availability, and is costly and time-consuming. Clinical prediction models thus are absolutely necessary in screening patients with OSA. Furthermore, the performance of the published prediction formulas is not satisfactory for Chinese populations. The aim of this study was to develop and validate a simple and accurate prediction system for the diagnosis of moderate to severe OSA by integrating an expert-based feature extraction technique with decision tree algorithms which have automatic feature selection capability in screening the moderate to severe OSA cases in Taiwan. Moreover, the backward stepwise multivariable logistic regression model and four other decision tree algorithms were also employed for comparison. The results showed that the proposed best prediction formula, with an overall accuracy reaching to 96.9 % in sensitivity = 98.2 % and specificity = 93.2 %, could present a good tool for screening moderate and severe Taiwanese OSA patients who require further PSG evaluation and medical intervention. Results also indicate that the proposed best prediction formula is simple, accurate, and reliable, and outperforms all the other prediction formulae considered in the present study. The proposed clinical prediction formula derived from three non-invasive features (Sex, Age, and AveSBP) may help prioritize patients for PSG studies as well as avoid a diagnosis of PSG in subjects who have a low probability of having the disease.

Developing screening services for colorectal cancer on Android smartphones.

INTRODUCTION: Colorectal cancer (CRC) is an important health problem in Western countries and also in Asia. It is the third leading cause of cancer deaths in both men and women in Taiwan. According to the well-known adenoma-to-carcinoma sequence, the majority of CRC develops from colorectal adenomatous polyps. This concept provides the rationale for screening and prevention of CRC. Removal of colorectal adenoma could reduce the mortality and incidence of CRC. Mobile phones are now playing an ever more crucial role in people's daily lives. The latest generation of smartphones is increasingly viewed as hand-held computers rather than as phones, because of their powerful on-board computing capability, capacious memories, large screens, and open operating systems that encourage development of applications (apps). SUBJECTS AND METHODS: If we can detect the potential CRC patients early and offer them appropriate treatments and services, this would not only promote the quality of life, but also reduce the possible serious complications and medical costs. In this study, an intelligent CRC screening app on Android™ (Google™, Mountain View, CA) smartphones has been developed based on a data mining approach using decision tree algorithms. For comparison, the stepwise backward multivariate logistic regression model and the fecal occult blood test were also used. RESULTS: Compared with the stepwise backward multivariate logistic regression model and the fecal occult blood test, the proposed app system not only provides an easy and efficient way to quickly detect high-risk groups of potential CRC patients, but also brings more information about CRC to customer-oriented services. CONCLUSIONS: We developed and implemented an app system on Android platforms for ubiquitous healthcare services for CRC screening. It can assist people in achieving early screening, diagnosis, and treatment purposes, prevent the occurrence of complications, and thus reach the goal of preventive medicine.