A Machine Learning Approach for Continuous Mining of Nonidentifiable Smartphone Data to Create a Novel Digital Biomarker Detecting Generalized Anxiety Disorder: Prospective Cohort Study.

BACKGROUND: Anxiety is one of the leading causes of mental health disability around the world. Currently, a majority of the population who experience anxiety go undiagnosed or untreated. New and innovative ways of diagnosing and monitoring anxiety have emerged using smartphone sensor-based monitoring as a metric for the management of anxiety. This is a novel study as it adds to the field of research through the use of nonidentifiable smartphone usage to help detect and monitor anxiety remotely and in a continuous and passive manner. OBJECTIVE: This study aims to evaluate the accuracy of a novel mental behavioral profiling metric derived from smartphone usage for the identification and tracking of generalized anxiety disorder (GAD). METHODS: Smartphone data and self-reported 7-item GAD anxiety assessments were collected from 229 participants using an Android operating system smartphone in an observational study over an average of 14 days (SD 29.8). A total of 34 features were mined to be constructed as a potential digital phenotyping marker from continuous smartphone usage data. We further analyzed the correlation of these digital behavioral markers against each item of the 7-item Generalized Anxiety Disorder Scale (GAD-7) and its influence on the predictions of machine learning algorithms. RESULTS: A total of 229 participants were recruited in this study who had completed the GAD-7 assessment and had at least one set of passive digital data collected within a 24-hour period. The mean GAD-7 score was 11.8 (SD 5.7). Regression modeling was tested against classification modeling and the highest prediction accuracy was achieved from a binary XGBoost classification model (precision of 73%-81%; recall of 68%-87%; F1-score of 71%-79%; accuracy of 76%; area under the curve of 80%). Nonparametric permutation testing with Pearson correlation results indicated that the proposed metric (Mental Health Similarity Score [MHSS]) had a colinear relationship between GAD-7 Items 1, 3 and 7. CONCLUSIONS: The proposed MHSS metric demonstrates the feasibility of using passively collected nonintrusive smartphone data and machine learning-based data mining techniques to track an individuals' daily anxiety levels with a 76% accuracy that directly relates to the GAD-7 scale.

Co-prescription patterns of cardiovascular preventive treatments: a cross-sectional study in the Aragon worker' health study (Spain).

OBJECTIVES: To identify cardiovascular disease (CVD) preventive treatments combinations, among them and with other drugs, and to determine their prevalence in a cohort of Spanish workers. DESIGN: Cross-sectional study. SETTING: Aragon Workers' Health Study (AWHS) cohort in Spain. PARTICIPANTS: 5577 workers belonging to AWHS cohort. From these subjects, we selected those that had, at least, three prescriptions of the same therapeutic subgroup in 2014 (n=4605). PRIMARY AND SECONDARY OUTCOME MEASURES: Drug consumption was obtained from the Aragon Pharmaceutical Consumption Registry (Farmasalud). In order to know treatment utilisation, prevalence analyses were conducted. Frequent item set mining techniques were applied to identify drugs co-prescription patterns. All the results were stratified by sex and age. RESULTS: 42.3% of men and 18.8% of women in the cohort received, at least, three prescriptions of a CVD preventive treatment in 2014. The most prescribed CVD treatment were antihypertensives (men: 28.2%, women 9.2%). The most frequent association observed among CVD preventive treatment was agents acting on the renin-angiotensin system and lipid-lowering drugs (5.1% of treated subjects). Co-prescription increased with age, especially after 50 years old, both in frequency and number of associations, and was higher in men. Regarding the association between CVD preventive treatments and other drugs, the most frequent pattern observed was lipid-lowering drugs and drugs used for acid related disorders (4.2% of treated subjects). CONCLUSIONS: There is an important number of co-prescription patterns that involve CVD preventive treatments. These patterns increase with age and are more frequent in men. Mining techniques are a useful tool to identify pharmacological patterns that are not evident in the individual clinical practice, in order to improve drug prescription appropriateness.

A multicenter study for the evaluation of the reference interval for TSH in Italy (ELAS TSH Italian Study).

Background The aims of this study were: (1) to calculate reliable thyroid stimulating hormone (TSH) reference intervals using laboratory databases; (2) to evaluate the relationship between TSH, sex and age values in different large Italian populations. Methods The TSH values stored in the laboratory information system of clinical laboratories of four Italian city hospitals, including 146,801 TSH measurements (with the respective age and sex data of individuals) were taken in consideration. Assuming a log-normal distribution, to log-transformed TSH values were applied the Dixon's iterative principle in order to exclude the outliers. At the end of this iterative process 142,821 log-transformed TSH results remained. The four clinical laboratories measured serum TSH concentrations using the same TSH immunoassay method (Access TSH 3rd IS, using UniCel DxI platform). Results The TSH reference interval calculated in the present study (0.362-5.280 mIU/L) is similar to that suggested by the manufacturer for the Access TSH 3rd IS assay (0.45-5.33 mIU/L). TSH values in females were significantly higher than in males (females: mean=2.06 mIU/L; standard deviation [SD]=1.26 mIU/L; n=101,243; males: mean=1.92 mIU/L; SD=1.19 mIU/L; n=41,578; p<0.0001). Moreover, a negative linear relationship was observed between TSH throughout all interval age values (from 0 to 105 years). Conclusions The results of the present multicenter study confirm that data mining techniques can be used to calculate clinically useful reference intervals for TSH. From a pathophysiological point of view, our results suggest that some Northern populations of Italy might still suffer some harmful effects on the thyroid gland due to mild to moderate iodine intake deficiency. Specific clinical trials are needed to confirm these results.

Drug-induced gingival hyperplasia: a retrospective study using spontaneous reporting system databases.

BACKGROUND: Drug-induced gingival hyperplasia (DIGH) causes problems with chewing, aesthetics, and pronunciation, and leads to the deterioration of the patient's quality of life (QOL). Thus, the aim of this study was to evaluate the incidence of DIGH using spontaneous reporting system (SRS) databases. METHODS: We analyzed reports of DIGH from SRS databases and calculated the reporting odds ratios (RORs) of suspected drugs (immunosuppressants, calcium channel blockers, and anticonvulsants). The SRS databases used were the US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) and the Japanese Adverse Drug Event Report (JADER) database. With the data, we evaluated the time-to-onset profile and the hazard type using the Weibull shape parameter (WSP). Furthermore, we used the association rule mining technique to discover undetected relationships such as possible risk factors. RESULTS: The FAERS contained 5,821,716 reports. The RORs (95% confidence interval: CI) for cyclosporine, everolimus, sirolimus, mycophenolate mofetil, amlodipine, nifedipine, carbamazepine, clobazam, levetiracetam, phenobarbital, phenytoin, primidone, topiramate, and valproic acid, were 39.4 (95% CI: 30.3-51.2), 4.2 (1.7-10.0), 6.6 (2.5-17.7), 13.1 (7.2-23.2), 94.8 (80.0-112.9), 57.9 (35.7-94.0), 15.1 (10.3-22.3), 65.4 (33.8-126.7), 6.5 (3.6-11.8), 19.7 (8.8-44.0), 65.4 (52.4-82.9), 56.5 (21.1-151.7), 2.9 (1.1-7.7), and 17.5 (12.6-24.4), respectively. The JADER database contained 430,587 reports. The median time-to-onset of gingival hyperplasia values for immunosuppressants, calcium channel blockers, and anticonvulsants use were 71, 262, and 37 days, respectively. Furthermore, the 95% CI of the WSP ß for anticonvulsants was over and excluded 1, which meant that they were wear-out failure type. CONCLUSIONS: Our results suggest that DIGH monitoring of patients administered immunosuppressants, calcium channel blockers, or anticonvulsants is important. We demonstrated the potential risk of DIGH following the long-term use of calcium channel blocker over approximately 260 days. Based on the results of the association rule mining approach, patients with intellectual disability who are administered phenytoin should be monitored carefully. We recommend that patients who experience symptoms related to DIGH should be closely monitored.

Exploration of the Medication Regularity of Traditional Chinese Medicine for Obesity Based on Text Mining Techniques / 世界科学技术-中医药现代化

This study aimed at analyzing the medication regularity based on differentiation in traditional Chinese medicine (TCM) for losing weight using text mining technique.All the references over losing weight were retrieved in CNKI,Wangfang Database,VIP Database and Pubmed.The drugs from the references were classified in accordance with drug property,drug flavor,channel tropism and drug efficacy.Frequency and constituent ratio of a single drug in TCM prescriptions for losing weight were put into analysis using chi square test and factor analysis to find out the medication regularity.It was found that the properties of TCM drugs in the prescriptions contained both cold and warm,while the flavors of the drugs involved pungent,sweet and light.The channel tropism of the drugs mainly belonged to spleen meridian,liver meridian,stomach meridian and lung meridian.They were mostly tonic,relieving,blood-activating,qi regulating,inhibiting-damp and antipyretic drugs.Through factor analysis we found that the common formula compatibilities were concluded as:cassia seed,lotus leaf,hawthorn,salvia miltiorrhiza,polygonum cuspidatum and radix polygonum multiflorum;capillary artemisia,epimedium herb,stephania tetrandra and ligusticum wallichii;dried tangerine peel,pinellia ternata and poria cocos;plantain seed,pericarpium arecae and selfheal;paeonia lactiflora,angelica sinensis,scutellaria baicalensis and ligusticum wallichii;poria cocos,cassia twig,atractylodes and glycyrrhiza;immature bitter orange and bark of magnolia;radix bupleuri,lycium chinensis and jujube;Chinese yam and coix seed;and astragalus,pueraria lobata and polygonatum.In conclusion,formula compatibility mainly combined syndrome differentiation with disease differentiation for the treatment of obesity in clinic,using the drugs belonging to liver meridian,spleen meridian,stomach meridian and lung meridian with the flavors of sweet,bitterness or pungent and the nature of both warm and cold.

Data Mining-based Analysis of Regularities in Acupoint Prescriptions for Stroke Disorders in Ancient Books / 上海针灸杂志

Objective To summarize regularities in acupoint prescriptions for acupuncture treatment of stroke disorders by using data mining techniques to analyze 150 ancient acupuncture books from the Han dynasty to the end of Qing dynasty. Method A standard database and a ancient book database were established by a manual entry and proofreading method. The sentences containing stroke disorders, acupoints, meridians and needling moxibustion method in the ancient books were selected according to the key words in the standard database using a variety of data mining techniques and a regular relation to establish corresponding disease, acupoint, needling moxibustion method and meridian databases. They were classified into two types: prescriptions for stroke disorders and acupoint indications containing stroke disorders. The results from selection were reviewed manually and standardized. A multilevel analysis of regularities in acupuncture prescriptions for stroke disorders in the ancient books was performed by statistical analysis and apriori algorithm on association rules with support degree and confidence level. Result There were a total of 536 records on acupuncture point prescription for stroke disorders. Single acupoint prescriptions accounted for 34.89% of the total number of prescriptions. The single acupoint that was used most frequently was Baihui(GV20). Thedouble acupoints that were combined most frequently were Quchi(LI11) and Baihui. In three acupoint combination, Quchi and Fengshi(GB31) were separately combined most frequently with Baihui, Jianyu(LI15) and Xuanzhong(GB39). The three meridians that were used most frequently were the Du Meridian, the Large Intestine Meridian of Hand-Yangming and the Gallbladder Meridian of Foot-Shaoyang. The two meridians that were combined most frequently were the Gallbladder Meridian of Foot-Shaoyang and the Stomach Meridian of Foot-Yangming. Crossing points and Five-Shu points were selected most frequently in specific points. He-Sea points were used most frequently in Five-Shu points. Conclusion The prescriptions for stroke disorders in ancient acupuncture books focus on selecting the confluence of all yang meridians and yang meridian points. Specific points are the important composition of the prescriptions. Crossing points and He-Sea points are mainly used in specific points.

Exploration on Medication Regularity of Epidemic Encephalitis B with Text Mining Technique / 中国中医药信息杂志

Objective To explore the clinical regularity of drugs on epidemic encephalitis B with text mining technique. Methods The data set on epidemic encephalitis B was downloaded from CBM database. The regularities of Chinese herbal medicine, Chinese patent medicine (CPM), western medicine and the combination of CPM and western medicine on epidemic encephalitis B were mined by data slicing algorithm. The results are showed visually with Cytoscape 2.8 software. Results The main function of herbal pieces was focused on clearing heat and removing toxicity. The herbal pieces such as Rehmanniae Radix Isatidis Raxis, Isatids Folium, Paeoniae Radis rubra and Scutellariae Radix were most often used. Angong Niuhuang Wan and Qingkailing Injection were usually used as CPM. For western medicine, antiviral drugs such as interferon and ribavirin were often used, and they were often used together with brain cells nutrition medicine, hormone, immunopotentiator and so on, and they were also frequently used together with CPM such as Angong Niuhuang Wan and Qingkailing Injection. Conclusion Text mining approach provides an important method in exploring the medication regularity for diseas, and provide refenrence for clinical use.

Exploring the associated rules of traditional Chinese medicine and western medicine on urinary tract infection with text mining technique / 中国医师进修杂志

Objective To explore the associated rules between western medicine and traditional Chinese medicine (TCM) on urinary tract infection (UTI) with text mining technique.Methods The data set on UTI was downloaded from CBM database.The regularities of Chinese patent medicines (CPM),western medicines and the combination of CPM and western medicines on UTI were mined out by data slicing algorithm.The results were showed visually with Cytoscape2.8 software.Results The main function of CPM was focused on clearing heat and removing toxicity,promoting diuresis and relieving stranguria.For western medicine,antibacterial agents was often used and it was also frequently used together with CPM such as Sanjinpian.Conclusions Text mining approach provides an important method in the summary of the application regularity for disease in both TCM and western medicine.