Background: Methylphenidate, a stimulant used to treat attention deficit hyperactivity disorder, has the potential to be used nonmedically, such as for studying and recreation. In an era when many people actively use social networking services, experience with the nonmedical use or side effects of methylphenidate might be shared on Twitter.
Objective: The purpose of this study was to analyze tweets about the nonmedical use and side effects of methylphenidate using a machine learning approach.
Methods: A total of 34,293 tweets mentioning methylphenidate from August 2018 to July 2019 were collected using searches for “methylphenidate” and its brand names. Tweets in a randomly selected training dataset (6860/34,293, 20.00%) were annotated as positive or negative for two dependent variables: nonmedical use and side effects. Features such as personal noun, nonmedical use terms, medical use terms, side effect terms, sentiment scores, and the presence of a URL were generated for supervised learning. Using the labeled training dataset and features, support vector machine (SVM) classifiers were built and the performance was evaluated using F1 scores. The classifiers were applied to the test dataset to determine the number of tweets about nonmedical use and side effects.
Results: Of the 6860 tweets in the training dataset, 5.19% (356/6860) and 5.52% (379/6860) were about nonmedical use and side effects, respectively. Performance of SVM classifiers for nonmedical use and side effects, expressed as F1 scores, were 0.547 (precision: 0.926, recall: 0.388, and accuracy: 0.967) and 0.733 (precision: 0.920, recall: 0.609, and accuracy: 0.976), respectively. In the test dataset, the SVM classifiers identified 361 tweets (1.32%) about nonmedical use and 519 tweets (1.89%) about side effects. The proportion of tweets about nonmedical use was highest in May 2019 (46/2624, 1.75%) and December 2018 (36/2041, 1.76%).
Conclusions: The SVM classifiers that were built in this study were highly precise and accurate and will help to automatically identify the nonmedical use and side effects of methylphenidate using Twitter.
Methylphenidate is a stimulant that is widely used for treating attention deficit hyperactivity disorder (ADHD) . It was approved for use in children and adolescents, and recently for adult ADHD, in several countries including the United States [ ]. The use of methylphenidate is increasing worldwide [ ]. Although methylphenidate is considered safe to use when taken as prescribed, it does have the potential for abuse because of its focus-enhancing, appetite-reducing, and euphoric effects [ , ]. In particular, more and more students are taking methylphenidate for academic purposes, calling it a “smart drug” or “study drug” [ - ]. A previous systematic review of 21 studies showed that 5%-9% of grade school- and high school-age children, as well as 5%-35% of college-age students, misused stimulants such as methylphenidate and amphetamines [ ]. In addition, children and adolescents take methylphenidate to stay up for parties and experience euphoria [ , ]. Due to the potential abuse of methylphenidates, many countries classify and control the drug legally [ , ].
Children and adult ADHD patients can benefit from the therapeutic effects of methylphenidate with few side effects when the drug is used as prescribed . However, studies have shown that children and adolescents who use methylphenidate to treat ADHD have a 60% higher risk of sleep disorders and a 266% higher risk of loss of appetite than those in control groups [ ]. In addition, children and adolescents with ADHD often have comorbid mood disorders and anxiety disorders, and the use of stimulants such as methylphenidate could exacerbate these comorbidities [ ]. Sometimes methylphenidate can cause hallucinations and delusions [ , ]. The rates of adverse drug reactions to methylphenidate, including agitation, irritability, and elevated heart rate, increase when it is abused [ , , ]. A total of 40% of 394 toxic exposures of methylphenidate reported in Denmark involved recreational use [ ]. Central nervous system and constitutional symptoms, such as anorexia, fatigue, and insomnia, were reported in 263 out of 323 cases (81.4%), and cardiovascular symptoms, such as arrhythmias, hypertension, and myocardial infarctions, were reported in 227 out of 323 (70.3%) of the symptomatic cases [ ]. The number of emergency room visits due to nonmedical use of ADHD medications nearly doubled between 2005 and 2010, from 5085 to 9181 [ ]. Thus, the abuse of methylphenidate has become a public health problem.
It is important to know the state of abuse and side effects of stimulants. Most studies of methylphenidate abuse have been conducted using surveys . The survey method is appropriate for investigating the status and motivation of nonmedical use; however, it is limited by efforts to conceal abuse, and some potential subjects may decline to participate in the study because of the fear of being discovered [ ]. Postmarketing surveillance studies using spontaneous reporting systems, such as the US Food and Drug Administration Adverse Event Reporting System, are suitable for analyzing the incidence and types of side effects caused by the use of methylphenidates but are limited in their ability to assess the current state of methylphenidate use.
As a new means of investigation, social networking services (SNSs) have begun to get attention in overcoming such challenges. Nowadays, it has become common to share one's thoughts, search for opinions, and interact with people with similar ideas through SNSs. In contrast to other SNSs, Twitter delivers most of its content as text rather than images, and many tools have been developed to analyze the content or emotions implicit in tweets. Consequently, it is relatively easy to analyze Twitter users' experiences or thoughts on a particular topic. Health researchers are increasingly using Twitter to analyze content about various topics (56%), as well as for surveillance (26%), engagement (14%), subject recruitment (7%), intervention (7%), and network analysis (4%) . From 2010 to 2015, the number of health researchers using Twitter increased almost 20-fold [ ]. Some studies have been conducted to analyze tweets about sentiment toward marijuana or tobacco smoking [ , ]. Recently, studies have been conducted that used machine learning to analyze Twitter’s big data repository. A study developed a classification program that could automatically detect opioid users from Twitter [ ]. Twitter and other SNSs are mainly utilized by younger users, and these groups are more likely to be diagnosed with ADHD or to abuse methylphenidate [ ]. The aim of this study was to analyze tweets about nonmedical use and side effects of methylphenidate using a machine learning approach.
The steps in this study were conducted in the following order: tweet collection, manual annotation, feature generation, supervised learning, and classification of the test dataset (see). The study was exempted from Institutional Review Board review (201909-HR-067-01).
Tweets mentioning methylphenidate from August 2018 to July 2019 were collected using the Twitter premium search application programming interface and Python, version 3.7.4 (Python Software Foundation). The following search terms for methylphenidate and its brand names were used: “methylphenidate,” “Aptensio,” “Biphentin,” “Concerta,” “Daytrana,” “Equasym,” “Jornay,” “Medikinet,” “Metadate,” “Methylin,” “Quillichew,” “Quillivant,” “Ritalin,” and “Rubifen.” This study did not cover the drug Adhansia because it was only approved in February 2019. Retweets were also collected if the user added their own text to an original tweet that contained the search terms. Duplicate tweets were removed, and only tweets written in English were used.
Among the collected tweets (N=34,293), 6860 (20.00%) were randomly selected as a training dataset. First, two annotators manually identified tweets mentioning first-hand experience. Tweets about drugs other than methylphenidate, song lyrics, humor, news, study results, or someone else’s experience were annotated as non-first-hand experience . Second, tweets about first-hand experience were classified as nonmedical use, side effects, and other. Tweets could be classified as pertaining to both nonmedical use and side effects. Finally, tweets were labeled positive and negative for two dependent variables: nonmedical use and side effects. Due to the nature of the short length of the text, it was often difficult to determine whether the drug was used nonmedically. In such cases, tweet threads or past tweets were checked to determine whether the user had ever been diagnosed with a condition that required methylphenidate. Interannotator agreement was assessed by Cohen kappa values [ , ], and any disagreements were resolved by discussion among psychiatrists.
Several features for supervised learning were generated: personal noun, nonmedical use terms, medical use terms, side effect terms, sentiment scores, and the presence of a URL (see).
|Feature and subfeature||Included terms|
|First person||i, i’, my, me, mine, myself, im, iam|
|Second person||you, you’, your, yours, yourself, ur|
|Third person||he, he’, his, him, himself, she, she’, her, hers, herself, they, they’, their, theirs, them, themselves|
|Others||boy, boyfriend, child, children, daughter, friend, girl, girlfriend, husband, kid, son, wife|
|Nonmedical use terms|
|General terms||abus, misus|
|Alternative motives||allnight, assign, clean, colleg, cram, diet, essay, exam, examin, final, focus, highschool, homework, loss, midnight, midterm, nighter, overnight, paper, paperwork, parti, project, quiz, recreat, school, shift, studi, studyin, test, work, write|
|Overdose||double, extra, overdos, overus, pop|
|Alternative route of administration||crush, inhal, inject, rail, sniff, sniffin, snort, snortin|
|Seeking||need, want, wish|
|Obtaining||buy, sell, share, steal, trade|
|Coingestion||alcohol, beer, bird, booz, bull, caffeine, cocain, coffe, coke, crack, crystal, energi, energydrink, espresso, heroin, lsd, marijuana, monster, pot, redbul, seed, shot, tequila, vodka, weed, wine, xtc|
|Medical use terms||addadhd, adhd, defici, diagnos, diagnosis, disord, narcolepsy, narcolept, prescribe, prescript|
|Side effect terms|
|General terms||side, sideeffect, advers|
|Loss of appetite||anorexia, appetit, ate, eat, eaten, eatin, food, hungry, lbs, meal, skinni, slim, starv, thin, underweight, weight|
|Sleep problems||asleep, awak, insomnia, insomniac, sleep, sleepi, sleepless, slept, tire|
|Psychiatric problems||anxieti, anxious, depress, jitter, jitteri, nervous, obsess, panic, restless, shaki, shakin, tens, tension, worri, zombie|
|Heart problems||beat, heart, heartbeat, heartrat, palpit|
|Gastrointestinal problems||diarrhea, dri, nausea, nauseat, nauseous, stomach, throw, thrown, vomit|
|Neurological problems||dizzi, head, headach, lighthead, migrain|
|Sweating||hot, sweat, sweatin|
|Eye problems||blurri, vision, visual|
|Presence of a URL||N/A|
The personal noun feature was generated to identify tweets mentioning first-hand experience. Personal nouns were grouped into one of four categories: first person, second person, third person, and others. Others included terms that could be used to describe someone else’s experience. Due to frequent nonstandard grammatical usage on Twitter, common modifications such as “im,” “iam,” and “ur” were also included.
Nonmedical Use Terms and Medical Use Terms
Prior to generating the features of nonmedical use terms, natural language processing was performed using the tm package from R, version 3.6.1 (The R Foundation), and RStudio. First, numbers, punctuation, and stop words, such as “the,” “is,” “and,” etc, were removed from the text of the tweets. Later, the tweet text was divided by word unit. Words were converted to their stems using the stemDocument function from R, and the frequency of word appearance in each tweet was described in a term-document matrix.
The counts of nonmedical use terms in individual tweets was used as a feature and this feature included seven subfeatures: general terms, alternative motives, overdose, alternative route of administration, seeking, obtaining, and coingestion. Terms were selected based on similar studies [, ]. Further, words related to nonmedical use were added by comparing words that appeared at a frequency of 5% or higher in tweets annotated as nonmedical use or negative. The counts of medical use terms were used to exclude medical use of methylphenidate for the treatment of ADHD or narcolepsy.
Side Effect Terms
The counts of side effect terms were generated as a feature after natural language processing as described above. Side effect terms included general terms, loss of appetite, sleep problems, psychiatric problems, heart problems, gastrointestinal problems, neurological problems, sweating, and eye problems. As in the selection of nonmedical terms, the terms included were added by reference to previous studies  or by comparing the words that appeared in the training dataset.
Sentiment scores were used as a feature because users often write polarized sentimental words when mentioning drug abuse or side effects. Sentiment scores were calculated by adding the number of positive words (each counting as +1) and the number of negative words (each counting as -1) appearing in a tweet. The Liu and Hu opinion lexicon dictionary, which contains 6800 positive and negative words in the English language, was used for sentiment analysis . Some negative words, such as “wtf,” were added to the dictionary.
Presence of a URL
This feature was created to identify retweets containing the user’s content or link to another website. Links to other websites were usually news or study results.
A support vector machine (SVM) with a radial basis function kernel was trained to classify nonmedical use and side effects using the e1071 package from R and RStudio. Two parameters of the SVM—cost and gamma—were tuned to achieve a better performance. Because the training dataset had a very large number of negative samples, 10-fold cross-validation was performed on the training data, and inverse weights were assigned to positive and negative samples to compensate for the imbalance . Due to the imbalance in the data, the F1 score (ie, harmonic mean of precision and recall) was used instead of accuracy to measure the performance of the SVM classifier. Precision, recall, accuracy, and F1 score were calculated as follows:
Precision = TP / (TP + FP) (1)
Recall = TP / (TP + FN) (2)
Accuracy = (TP + TN) / (TP + TN + FP + FN) (3)
F1 = (2 × Precision × Recall) / (Precision + Recall) (4)
True positives (TP), false positives (FP), true negatives (TN), and false negatives (FN) were calculated by comparing annotated results and predicted results.
Classification of Test Dataset
The SVM classifier separated the test dataset into nonmedical use or side effects and negative. Tweets about nonmedical use and side effects were counted. The number of tweets about nonmedical use of methylphenidate each month was determined to examine the correlation with the school term.
From August 2018 to July 2019, 36,578 tweets were collected using predetermined search terms. Tweets containing the words “Ritalin” and “Concerta” were the most frequent (27,635/36,578, 75.55%, and 5485/36,578, 15.00%, respectively). In total, 34,293 nonduplicated tweets were ultimately selected: 6860 (20.00%) in the training dataset and 27,433 (79.99%) in the test dataset.
Among the 6860 tweets in the training dataset, 2108 (30.73%) mentioned first-hand experience, including 356 about nonmedical use (5.19%) and 379 about side effects (5.52%). A total of 20 tweets (0.29%) were annotated as pertaining to both nonmedical use and side effects. Cohen kappa values were .73 and .75 for nonmedical use and side effects, respectively, which means there was substantial agreement between the two annotators.
The classification performance of SVM classifiers, expressed as an F1 score, was 0.547 for nonmedical use and 0.733 for side effects (seeand ). The low recall of the SVM classifier for nonmedical use, despite its high precision, was responsible for its low F1 score. Each feature contributed to improvement of SVM classifiers (see and ). With the exception of nonmedical use and side effect terms, F1 scores were the lowest when the feature personal noun was excluded.
|SVM classifier||F1 score for nonmedical use||F1 score for negative||Precision||Recall||Accuracy|
|Without nonmedical use terms||0.145||0.975||0.933||0.079||0.952|
|Without medical use terms||0.506||0.982||0.925||0.348||0.965|
|Without personal noun||0.233||0.976||0.857||0.135||0.954|
|Without sentiment scores||0.420||0.979||0.833||0.281||0.960|
|Without a URL||0.526||0.982||0.923||0.368||0.966|
|SVM classifier||F1 score for side effects||F1 score for negative||Precision||Recall||Accuracy|
|Without side effect terms||0.316||0.976||0.880||0.193||0.954|
|Without personal noun||0.388||0.978||0.887||0.248||0.957|
|Without sentiment scores||0.571||0.982||0.918||0.414||0.966|
|Without a URL||0.722||0.987||0.922||0.594||0.975|
From the test dataset (n=27,433), 361 tweets (1.32%) about the nonmedical use and 519 tweets (1.89%) about the side effects of methylphenidate were identified using SVM classifiers. A total of 21 tweets (0.08%) were classified as pertaining to both nonmedical use and side effects. Examples of tweets, paraphrased to ensure anonymity, classified as nonmedical use included “When I was young I snorted my Concerta only for the head rush” and “Time to pop the Ritalin I been keeping.” Paraphrased tweets classified as side effects included “Worst 9 days of my life. I thought Ritalin would calm the anxiety part of the ADHD. But it makes me a short-fused angry psycho who burst into tears spontaneously” and “Seizures, hallucinations, paranoia is all Ritalin brought me. The side effects still bother me.”
The monthly proportion of tweets about the nonmedical use of methylphenidate is shown in. The proportion was highest in May 2019 (46/2624, 1.75%) and December 2018 (36/2041, 1.76%), which are the final exam periods in the United States.
This study was conducted to analyze tweets about the nonmedical use and side effects of methylphenidate. Because there were more than 30,000 tweets mentioning methylphenidate written in a year, it was difficult to classify them manually; therefore, we used the SVM machine learning approach. Similar stimulants, such as Adderall, mixed amphetamine salts, have been studied before [, ]. An early study using Twitter measured the co-occurrence of nonmedical or side effect terms among 213,633 tweets mentioning Adderall [ ]. However, not all Adderall tweets referenced first-hand experience; as the author of that study mentioned, the analysis included 5169 song lyrics [ ]. Another recent study automatically detected tweets related to the nonmedical use of Adderall using an SVM approach [ ]. However, the SVM classifier had a poor performance: F1 score of 0.46, precision of 0.41, and recall of 0.51 [ ]. Our study is the first to analyze tweets about the nonmedical use and side effects of methylphenidate and has two main advantages: training first-hand experience and better performance.
In the annotation process, nonmedical use of methylphenidate was identified in 5% of the training dataset, lower than in two previous studies about Adderall (12.9% and 22.6%) [, ]. This may be due to the popularity of Adderall. In a survey of 4580 college students, three-quarters of those who had engaged in nonmedical use of stimulants over the past year had used Adderall, and one-quarter had taken methylphenidate [ ]. In another survey, 54.2% of respondents who abused stimulants used Adderall versus 15% who used methylphenidate [ ]. Alternatively, this may be due to stringent standards used in our study: we included only first-hand experience and evaluated whether methylphenidate was administered for medical purposes based on past tweets from the users.
Nonmedical use and side effect terms improved the SVM classifiers the most. The inclusion of personal nouns in the model also significantly improved the classifier. Sentiment scores also contributed to the improvement of the classifier, although they did not capture users’ exact sentiments toward methylphenidate. The previous Adderall study included sentiment analysis in the SVM classifier, which slightly improved the F1 scores .
The F1 scores of SVM classifiers for nonmedical use and side effects were 0.547 and 0.733, respectively. SVM classifiers had low F1 scores due to low recall, which may have induced underestimation of nonmedical use and side effects in the test dataset (1.3% and 1.9%, respectively). In particular, the recall of nonmedical use was low because the classifier was built solely on the content of tweets, although the label was annotated by reviewing the users’ previous tweets to see if they had been diagnosed with ADHD. However, because precision was high, tweets classified as nonmedical use can be thought of as TP.
In May 2019 and December 2018, the US exams periods, the proportion of tweets related to nonmedical use of methylphenidate was highest. Tweet timing may differ from the time of administration because users sometimes write about past experiences. Nevertheless, higher ratios relative to the other periods may indicate frequent administration of methylphenidate during the exam period for the purpose of improving concentration. The increase in the number of tweets about stimulants in May 2019 and December 2018 was also reported in the Adderall studies [, ].
The final SVM classifiers had low recall, especially for tweets about nonmedical use. This is because some tweets were ambiguous and previous tweets were not considered in the SVM classifier. If user information or previous tweet information could be included as a feature, it might help to solve the problem of low recall. Another problem is scarcity of positive samples relative to negative samples, despite the use of several methods to resolve the problem of imbalance. The lack of positive data made it impossible to learn about individual side effects, such as sleep disorders and heart problems. Furthermore, the SVM classifier for side effects was not sufficient to detect new side effects because the terms corresponding to known representative side effects were used as features in the supervised learning process.
The frequency of tweets reported in this study does not imply the actual prevalence of nonmedical use or side effects of methylphenidate. First, the study only targeted users of Twitter, which restricts use by individuals under the age of 13 years. Second, the calculated percentage is the percentage of tweets, not the percentage of respondents, as in a general survey. Third, not all Twitter users who take methylphenidate will necessarily write tweets about the drug and among those who do, some may include the drug name and information about nonmedical use or side effects in separate tweets within the same thread. Finally, the study did not take into account various typos in the search terms or non-English tweets.
This study built SVM classifiers that helped to automatically identify the nonmedical use and side effects of methylphenidate from Twitter. The SVM classifiers had high precision and accuracy but low recall. Information available on Twitter is not available during the prescription process and cannot be identified through electronic medical records. Similar information can be obtained through surveys, as in previous studies, but research using Twitter has the advantage of saving time and cost required for a survey. Future studies should seek to apply this method to other social media platforms.
Conflicts of Interest
- Morton WA, Stockton GG. Methylphenidate abuse and psychiatric side effects. Prim Care Companion J Clin Psychiatry 2000 Oct;2(5):159-164 [FREE Full text] [CrossRef] [Medline]
- Label for CONCERTA (methylphenidate HCl) Extended-Release Tablets. 2017. URL: https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/021121s038lbl.pdf [accessed 2019-09-10]
- Jensen LS, Pagsberg AK, Dalhoff K. Methylphenidate misuse in adult patients and the impact of therapeutic use. Hum Exp Toxicol 2015 May;34(5):460-467. [CrossRef] [Medline]
- Clemow D. Misuse of methylphenidate. Curr Top Behav Neurosci 2017;34:99-124. [CrossRef] [Medline]
- Goldfield GS, Lorello C, Doucet E. Methylphenidate reduces energy intake and dietary fat intake in adults: A mechanism of reduced reinforcing value of food? Am J Clin Nutr 2007 Aug;86(2):308-315. [CrossRef] [Medline]
- Teter CJ, McCabe SE, Cranford JA, Boyd CJ, Guthrie SK. Prevalence and motives for illicit use of prescription stimulants in an undergraduate student sample. J Am Coll Health 2005;53(6):253-262. [CrossRef] [Medline]
- Teter CJ, McCabe SE, LaGrange K, Cranford JA, Boyd CJ. Illicit use of specific prescription stimulants among college students: Prevalence, motives, and routes of administration. Pharmacotherapy 2006 Oct;26(10):1501-1510 [FREE Full text] [CrossRef] [Medline]
- Barrett SP, Darredeau C, Bordy LE, Pihl RO. Characteristics of methylphenidate misuse in a university student sample. Can J Psychiatry 2005 Jul;50(8):457-461. [CrossRef] [Medline]
- White BP, Becker-Blease KA, Grace-Bishop K. Stimulant medication use, misuse, and abuse in an undergraduate and graduate student sample. J Am Coll Health 2006;54(5):261-268. [CrossRef] [Medline]
- Sussman S, Pentz MA, Spruijt-Metz D, Miller T. Misuse of "study drugs:" Prevalence, consequences, and implications for policy. Subst Abuse Treat Prev Policy 2006 Jun 09;1:15 [FREE Full text] [CrossRef] [Medline]
- Wilens TE, Adler LA, Adams J, Sgambati S, Rotrosen J, Sawtelle R, et al. Misuse and diversion of stimulants prescribed for ADHD: A systematic review of the literature. J Am Acad Child Adolesc Psychiatry 2008 Jan;47(1):21-31. [CrossRef] [Medline]
- Storebø OJ, Ramstad E, Krogh HB, Nilausen TD, Skoog M, Holmskov M, et al. Methylphenidate for children and adolescents with attention deficit hyperactivity disorder (ADHD). Cochrane Database Syst Rev 2015 Nov 25(11):CD009885. [CrossRef] [Medline]
- Pozzi M, Carnovale C, Mazhar F, Peeters GG, Gentili M, Nobile M, et al. Adverse drug reactions related to mood and emotion in pediatric patients treated for attention deficit/hyperactivity disorder: A comparative analysis of the US Food and Drug Administration Adverse Event Reporting System database. J Clin Psychopharmacol 2019;39(4):386-392. [CrossRef] [Medline]
- Hollis C, Chen Q, Chang Z, Quinn PD, Viktorin A, Lichtenstein P, et al. Methylphenidate and the risk of psychosis in adolescents and young adults: A population-based cohort study. Lancet Psychiatry 2019 Aug;6(8):651-658 [FREE Full text] [CrossRef] [Medline]
- Moran LV, Ongur D, Hsu J, Castro VM, Perlis RH, Schneeweiss S. Psychosis with methylphenidate or amphetamine in patients with ADHD. N Engl J Med 2019 Mar 21;380(12):1128-1138 [FREE Full text] [CrossRef] [Medline]
- Rabiner DL. Stimulant prescription cautions: Addressing misuse, diversion and malingering. Curr Psychiatry Rep 2013 Jul;15(7):375. [CrossRef] [Medline]
- Sinnenberg L, Buttenheim AM, Padrez K, Mancheno C, Ungar L, Merchant RM. Twitter as a tool for health research: A systematic review. Am J Public Health 2017 Jan;107(1):e1-e8. [CrossRef] [Medline]
- Cavazos-Rehg PA, Krauss M, Fisher SL, Salyer P, Grucza RA, Bierut LJ. Twitter chatter about marijuana. J Adolesc Health 2015 Feb;56(2):139-145 [FREE Full text] [CrossRef] [Medline]
- Myslín M, Zhu S, Chapman W, Conway M. Using Twitter to examine smoking behavior and perceptions of emerging tobacco products. J Med Internet Res 2013 Aug 29;15(8):e174 [FREE Full text] [CrossRef] [Medline]
- Fan Y, Zhang Y, Ye Y, Li X. Automatic opioid user detection from Twitter: Transductive ensemble built on different meta-graph based similarities over heterogeneous information network. In: Proceedings of the 27th International Joint Conference on Artificial Intelligence. 2018 Jul Presented at: 27th International Joint Conference on Artificial Intelligence; July 13-19, 2018; Stockholm, Sweden p. 3357-3363. [CrossRef]
- Meng H, Kath S, Li D, Nguyen QC. National substance use patterns on Twitter. PLoS One 2017;12(11):e0187691 [FREE Full text] [CrossRef] [Medline]
- Alvaro N, Conway M, Doan S, Lofi C, Overington J, Collier N. Crowdsourcing Twitter annotations to identify first-hand experiences of prescription drug use. J Biomed Inform 2015 Dec;58:280-287 [FREE Full text] [CrossRef] [Medline]
- Carletta J. Assessing agreement on classification tasks: The kappa statistic. Comput Linguist Assoc Comput Linguist 1996 Jun;22(2):249-254.
- McHugh ML. Interrater reliability: The kappa statistic. Biochem Med (Zagreb) 2012;22(3):276-282 [FREE Full text] [Medline]
- Hanson CL, Burton SH, Giraud-Carrier C, West JH, Barnes MD, Hansen B. Tweaking and tweeting: Exploring Twitter for nonmedical use of a psychostimulant drug (Adderall) among college students. J Med Internet Res 2013 Apr 17;15(4):e62 [FREE Full text] [CrossRef] [Medline]
- Hanson CL, Cannon B, Burton S, Giraud-Carrier C. An exploration of social circles and prescription drug abuse through Twitter. J Med Internet Res 2013 Sep 06;15(9):e189 [FREE Full text] [CrossRef] [Medline]
- Liu B, Hu M. University of Illinois at Chicago, Computer Science. 2004 May 15. Opinion mining, sentiment analysis, and opinion spam detection URL: https://www.cs.uic.edu/~liub/FBS/sentiment-analysis.html#lexicon [accessed 2019-08-30]
- Blagus R, Lusa L. Joint use of over- and under-sampling techniques and cross-validation for the development and assessment of prediction models. BMC Bioinformatics 2015 Nov 04;16:363 [FREE Full text] [CrossRef] [Medline]
- Sarker A, O'Connor K, Ginn R, Scotch M, Smith K, Malone D, et al. Social media mining for toxicovigilance: Automatic monitoring of prescription medication abuse from Twitter. Drug Saf 2016 Mar;39(3):231-240 [FREE Full text] [CrossRef] [Medline]
- Bright GM. Abuse of medications employed for the treatment of ADHD: Results from a large-scale community survey. Medscape J Med 2008 May 07;10(5):111. [Medline]
|ADHD: attention deficit hyperactivity disorder|
|FN: false negatives|
|FP: false positives|
|SNS: social networking service|
|SVM: support vector machine|
|TN: true negatives|
|TP: true positives|
Edited by G Eysenbach; submitted 02.10.19; peer-reviewed by A Sarker, JP Allem; comments to author 28.10.19; revised version received 08.01.20; accepted 27.01.20; published 24.02.20Copyright
©Myeong Gyu Kim, Jungu Kim, Su Cheol Kim, Jaegwon Jeong. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 24.02.2020.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.