Where Can I Buy Xanax Without a Prescription

J Med Internet Res. 2022 Aug; 22(viii): e17239.

Managing Illicit Online Pharmacies: Spider web Analytics and Predictive Models Written report

Monitoring Editor: Gunther Eysenbach

Hui Zhao, PhD, ¹ Sowmyasri Muthupandi, MSc,² and Soundar Kumara, PhD³

¹ Smeal College of Business, Pennsylvania Country University, Academy Park, PA, United States,

² PricewaterhouseCoopers, LLP, New York, NY, United States,

³ Section of Industrial and Manufacturing Engineering, Pennsylvania Land University, University Park, PA, United States,

Hui Zhao, Smeal Higher of Business, 419 Business Building, Pennsylvania Land Academy, University Park, PA, 16802, The states, Telephone: one 814 863 1940, Email: ude.usp@01zuh.

Hui Zhao

¹ Smeal College of Business, Pennsylvania State Academy, University Park, PA, United States,

Sowmyasri Muthupandi

^two PricewaterhouseCoopers, LLP, New York, NY, The states,

Soundar Kumara

³ Section of Industrial and Manufacturing Applied science, Pennsylvania State University, University Park, PA, United states,

Received 2022 Dec iii; Revisions requested 2022 Feb 10; Revised 2022 April xiv; Accustomed 2022 May 14.

Supplementary Materials

Multimedia Appendix i.

More details of the engagement and traffic source data.

GUID: 2A788F1F-A207-47C9-9788-E094F3DFAD00

Abstruse

Background

Online pharmacies accept grown significantly in recent years, from US $29.35 billion in 2014 to an expected United states $128 billion in 2023 worldwide. Although legitimate online pharmacies (LOPs) provide a channel of convenience and potentially lower costs for patients, illicit online pharmacies (IOPs) open the doors to unfettered access to prescription drugs, controlled substances (eg, opioids), and potentially counterfeits, posing a dramatic adventure to the drug supply chain and the health of the patient. Unfortunately, we know little about IOPs, and even identifying and monitoring IOPs is challenging considering of the big number of online pharmacies (at least 30,000-35,000) and the dynamic nature of the online channel (online pharmacies open and close down hands).

Objective

This written report aims to increment our understanding of IOPs through web data traffic analysis and propose a novel framework using referral links to predict and identify IOPs, the beginning step in fighting IOPs.

Methods

We starting time collected spider web traffic and engagement information to study and compare how consumers access and appoint with LOPs and IOPs. We then proposed a simple but novel framework for predicting the status of online pharmacies (legitimate or illicit) through the referral links between websites. Under this framework, nosotros developed 2 prediction models, the reference rating prediction method (RRPM) and the reference-based Yard-nearest neighbor.

Results

We institute that straight (typing URL), search, and referral are the three major traffic sources, representing more than 95% traffic to both LOPs and IOPs. It is alarming to see that direct represents the second-highest traffic source (34.32%) to IOPs. When tested on a data prepare with 763 online pharmacies, both RRPM and R2NN performed well, achieving an accuracy higher up 95% in their predictions of the status for the online pharmacies. R2NN outperformed RRPM in full performance metrics (accurateness, kappa, specificity, and sensitivity). On implementing the 2 models on Google search results for popular drugs (Xanax [alprazolam], OxyContin, and opioids), they produced an mistake rate of only vii.96% (R2NN) and 6.twenty% (RRPM).

Conclusions

Our prediction models use what we know (referral links) to tackle the many unknown aspects of IOPs. They accept many potential applications for patients, search engines, social media, payment companies, policy makers or government agencies, and drug manufacturers to aid fight IOPs. With scarce work in this area, we promise to help accost the current opioid crisis from this perspective and inspire future research in the critical area of drug condom.

Keywords: online pharmacy, spider web analytics, classification, illicit online pharmacies, online traffic analysis

Introduction

Online pharmacies (OPs) have grown tremendously in recent years, from US $29.35 billion in 2014 to an expected US $128 billion in 2023 globally, at an annual growth rate of 17.vii% [one]. Most consumers pursue OPs for lower prices [two,iii], convenience, and access to otherwise unavailable drugs, for example, recalled or on shortage [four,5]. All the same, there is bereft awareness of the prevalent illicit online pharmacies (IOPs), which are estimated to represent 67%-75% web-based drug merchants [6]. Although much work has been carried out to restrict prescription for the recent opioid crisis, many IOPs provide admission without prescription. IOPs provide unfettered access to prescription drugs and even controlled substances, leading to great concerns about substandard drugs, counterfeits, and supply concatenation integrity [seven,eight].

Fighting IOPs is critical in protecting patient safety as well as integrity of the drug supply chain. However, this is very challenging. First, in that location is low sensation of how to differentiate the legitimacy of OPs among consumers [9], and we still have much to learn about IOPs [6]. IOPs may look very similar to LOPs, and, unlike other consumer products, most consumers take no expertise in differentiating potentially substandard drugs even upon receiving them. Second, even identifying and tracking IOPs, the starting time stride in fighting IOPs, can be challenging because of the sheer calibration and the dynamic nature of the problem. Co-ordinate to Legitscript [6], there are xxx,000-35,000 online pharmacies, and about 20 new IOPs are created when many die on a daily footing. Fifty-fifty if IOPs can exist closed down (more difficult than we recall every bit many IOPs take their servers outside of the United States), they can hands pop up using different URLs (eg, 30,000-35,000 known OPs represent only 2000-3500 merchants [six]).

A few checking systems of OP status (legitimate or illicit) do exist but with limitations. Some of them are not recommended [x], including the Canadian International Pharmacy Association and Pharmacychecker, which have been criticized for not always classifying the OPs correctly. The two sources recommended by the Nutrient and Drug Administration are the National Association Board of Pharmacies (NABP) and Legitscript. Withal, both sources require consumers to have the initiative to await up the status of the pharmacies. Co-ordinate to a survey of 500 consumers from the United States, conducted by the Alliance for Safe Online Pharmacies, 95% do not know about the certification programs [9], let alone where to bank check the status of the OPs. Furthermore, in that location is no exhaustive database because of the same calibration and the dynamic nature of OPs.

This study aims to apply spider web analytics to better understand IOPs and to predict, identify and monitor IOPs using known information. Nosotros do this in 2 steps. Starting time, we conducted a traffic analysis based on web-nerveless data, which assesses the means through which LOPs and IOPs are accessed and how engaged the customers are with them. On the basis of the information from the first footstep, peculiarly through the assay of referrals data, in the 2d step, we proposed a novel framework to predict the condition (legitimate or illicit) of OPs based on the referral websites to them. Under this framework, we adult 2 easy-to-empathize prediction models, the referral-based K-nearest neighbor (RKNN) and the referral rating prediction method (RRPM), and tested them using a data gear up with 763 OPs. Nosotros then implemented the 2 methods on Google search results for 3 popular drugs: Xanax (alprazolam), OxyContin, and opioids. These methods accept many potential applications for consumers when shopping on the spider web and for other stakeholders to aid fight IOPs, as presented in detail in the Applications and Conclusions subsection of the Give-and-take section.

Methods

Data Sources

Nosotros obtained the basis truth list of LOPs and IOPs from the NABP (Legitscript was non available for the size of our sample). NABP provided a list of approximately yard IOPs and l LOPs. We filtered out many IOPs that stopped operations at the time of data collection. We then collected usage data (ie, traffic and engagement information) for the remaining OPs from Similarweb and obtained the structure data (ie, referrals and backlink information, detailed later) from SEMrush. As Similarweb does not have data for websites non in its database or whose traffic is besides low to monitor, this led the states to the final sample sizes for each of the databases in Table one. The kickoff 5 rows in the tabular array are the usage information, and the last row is the structure data. We collected data from Similarweb through web scraping using R. For SEMrush, we tried to collect the data manually (no crawling allowed for SEMrush). When that was incommunicable, we purchased the function from SEMrush (it sells different levels of functions through various priced accounts).

Table 1

Data sets and sample size.

Data set names	Legitimate pharmacies, n	Illicit pharmacies, northward	Total samples, n	Data collection period
Traffic sources data	30	127	157	Average over 4 months (October 2015-February 2016)
Engagement data	xxx	127	157	Average over 4 months (October 2015-February 2016)
State data	thirty	139	169	Average over four months (Oct 2015-February 2016)
Social media information	24	41	65	Average over four months (Oct 2015-Feb 2016)
Search information	xxx	sixty	90	Boilerplate over 4 months (October 2015-February 2016)
Referral data	50	713	763	September 2016

In Table 1, traffic sources provide the percent of the sources through which consumers access the OPs, that is, straight, search, referral, social media, brandish, and email (details later). Engagement data testify the extent of users' interest with the website (eg, the number of pages viewed and fourth dimension spent on the website). Country data provide the pct of traffic to OPs from different countries. Social media data refer to the proportion of traffic from 26 social media websites, such as Facebook, YouTube, and Google Plus. Search data provide the percentages of traffic resulting from organic or paid searches for OPs. An organic search, also called natural search, provides results by the search engine based on its relevance to the user's query. Paid search results are like advertisements, where the websites pay search engines to promote their web pages for particular keywords. Referral data provide the different referring websites to online pharmacies, their internet protocol addresses, and countries of origin.

OP Status Prediction Model

1 of the difficulties in predicting the status of an OP is that the proposed method and the data it uses need to exist something that cannot be hands manipulated by IOPs to affect future prediction results. To overcome this challenge, nosotros propose a novel structure-based framework that predicts the status based on the relationship among the referral websites. Basically, we expect that if a pharmacy is mainly reached from referral websites that by and large link/refer to illicit pharmacies, so this pharmacy is more than likely to be illicit. Figure one depicts an oversimplified sit-in of this idea and the links between referral websites.

Uncomplicated Sit-in of Links Between Referral Websites and online pharmacies.

To execute this idea, based on the footing truth listing of LOPs and IOPs from NABP, we identified all the websites referring to the OPs in the data set and nerveless the construction data, that is, referrals and data of the number of backlinks to each OP, where a backlink is a link from a website to some other website (eg, the OP here). These data, listed in Table 1 as the referral data, were and so used to train the prediction model. Tabular array 2 provides a snapshot of these data (the entries are the number of backlinks from a referral site j to a pharmacy i).

Table 2

Sit-in of our data set for the prediction model.

Pharmacy site, i	Referral site, j
	1	2	iii	...	j	...
1	5	0	3	…	16	…
2	ix	3	0	…	0	…
i	0	0	0	…	2	…
...	…	…	…	…	…	…

Figure 2 plots all our referral information. In the figure, the pink nodes are the IOPs, the dark-green nodes are the LOPs, and the blue nodes are their referral websites. This figure shows 2 interesting phenomena: (1) LOPs and IOPs are clearly separated past the referral websites directing to them (although some referral websites refer to both IOPs and LOPs), that is, IOPs tend to exist referred to by referring websites referring to other IOPs, and vice versa and (2) adept referral websites tend to cluster in groups referring to each other'south referred pharmacies, whereas bad referral websites scatter around (they refer to all kinds of pharmacies far and between). These 2 phenomena, especially the outset ane, confirms our bones thought of using the quality of the referral websites (ie, how much the referring websites refer to LOPs) to predict the status of the OPs.

Relationship among referral websites and LOPs and IOPs based on real data.The pink nodes are the IOPs, the greenish nodes are the LOPs and the blue nodes are their referral websites. LOPs: legitimate online pharmacies; IOPs: illicit online pharmacies.

Then, nosotros described in detail 2 prediction models that we developed based on this idea, that is, the RRPM and the RKNN.

RRPM

Permit represent the set of pharmacies, stand for the fix of referring websites, l_ij correspond the number of backlinks to chemist's shop i from referring website j, and y_i = ane (illicit) or 0 (licit) stand for the status of chemist's i. We are at present set to present the model.

Pace 1: Kickoff define the quality of a referral website j (M_j) based on its backlinks to legitimate and illicit pharmacies as

where represents the prepare of safe or legitimate pharmacies and represents the fix of rogue or illicit pharmacies.

Therefore, represent the number of LOPs, IOPs, and whatsoever OPs website j refers to, respectively. Information technology is easy to see that M_j is between –1 and 1 with M_j =−1 indicating that website j only refers to IOPs and M_j =1 indicating that website j only refers to LOPs.

Step 2: For pharmacy i whose status is to be predicted, calculate the reliability score (R_i) as

where is the total number of backlinks (referral websites) to pharmacy i. Note that information technology is possible that a given pharmacy to be predicted does not take any referral website from the training set. In this case, its R_i volition be indeterminate and R_i is fix to 0.

On the basis of our framework, we expect that the higher the R_i is, the more probable information technology is legitimate. For our prediction model, we fix a threshold T for the reliability score in a higher place which nosotros predicted the pharmacy to exist legitimate. In determining T, we considered a crucial factor, the sensitivity of the model, which measured the proportion of IOPs that were correctly identified equally such. Although predicting a pharmacy wrong in either mode is risky, for safety reasons, from the consumers' perspective, classifying an illicit pharmacy as legitimate may be more detrimental than classifying a legitimate chemist's as illicit. Taking this into consideration, we have the following:

Pace 3: Set up the threshold T from the training fix every bit

such that we volition allocate chemist's shop i as illicit if R_i<T and legitimate otherwise.

Notice that this is a very conservative threshold, pregnant that a pharmacy is highly unlikely to be an illicit one when it is predicted legitimate. This could injure the average accuracy of the model. Therefore, considering the average accuracy, i could select a different threshold. Another fashion we tried is to examination different threshold levels with the training set and choose the one with the highest accurateness. Later, nosotros reported the accuracy for both thresholds.

RKNN

In addition to RRPM, nosotros next adapted i of the established classification methods, One thousand-nearest neighbor (KNN), to our framework based on the referral links to develop another prediction model. KNN is a supervised learning model that classifies the samples in the test fix based on their proximity to the samples of unlike classes in the grooming set [11,12]. The primal to this method is defining proximity (similarity). Nosotros now incorporated our idea of the proposed framework into this definition.

Step one: Compute the Euclidean distance between the chemist's ten (the one whose status is to be predicted) and all the online pharmacies i with known status i=1,2,…,n, every bit

Note that the smaller the D_i is, the more than similar chemist's shop x is to chemist's shop i in terms of the referring websites directing to them.

Stride 2: Social club the online pharmacies in decreasing order with respect to D_i. Note downward the status of the top Grand pharmacies. According to the traditional KNN, the status/class of 10 is assigned to the more frequent status among the K pharmacies. Formally, let the number of legitimate pharmacies among the summit 1000 exist K_s and illicit ones exist K_r. We know that K_{due south} +K_r=Thousand. Allow

x volition exist predicted to be legitimate if R_x >0.v. R₁₀ is similar to the reliability score of ten, indicating the force of the prediction, with a higher R_x signifying a stronger prediction.

Like to KNN, the performance of the RKNN model varies for different values of Thousand. Obviously, a too high or besides low value of Grand may reduce the accurateness of the model. Nosotros tested Thousand=1, 2,…9 and reported the performance of the model for each value of Grand.

Results

Traffic and Use Analysis of the Online Pharmacies

Traffic sources of all websites are classified as direct, search, referral, social media, display, and email. Specifically, traffic obtained by users' direct typing in the URL of the website is classified as direct; search refers to the traffic coming from search engines such as Google, Bing, and Yahoo; traffic from links on other websites are accounted for equally referral; social indicates the traffic from social media such every bit Facebook and Twitter; brandish indicates the traffic from banner advertising; and E-mail indicates the traffic coming from links in email messages.

Table 3 shows the mean per centum of traffic from each source to the IOPs and LOPs in our traffic data set (the standard divergence is shown in Multimedia Appendix 1). According to Table 3, direct, search, and referral are the 3 major traffic sources, representing more than 95% traffic to both LOPs and IOPs. A loftier per centum of straight traffic indicates that the specific OP website is a powerful brand and users visiting the website know what they desire. Although LOPs are most accessed through direct traffic (42.48%), it is alarming to meet that direct traffic also represents the 2nd highest traffic source for IOPs (34.32%). This indicates that consumers who take previous experiences with IOPs (eg, from search, referrals) may become returning customers without knowing the aforementioned potential danger. Therefore, it is imperative to educate and alert patients, and curb people from using IOPs for the first time.

Tabular array 3

Mean percentages of dissimilar traffic sources to online pharmacies.

Traffic source	Legitimate online pharmacy (n=30), %	Illicit online pharmacy (north=127), %
Directly	42.v	34.3
Search	36.3	39.3
Referral	17.7	21.7
Social	1.iii	0.9
Email	ii.ii	0.6
Display	0	2.five

Previous research has indicated the presence of IOP contents on various social media sites [3]. Our data show that the average percentages of traffic through social media are less than 5% for both IOPs and LOPs, possibly because many OPs in our sample do not have traffic from social media. When we focused on the 24 LOPs and 41 IOPs from our sample that have substantial traffic from the 26 social media sites to conduct farther analysis, we institute that 92% (24/26) of the studied social media websites direct traffic to IOPs and just fifty% (xiii/26) of them direct traffic to LOPs. Although 42% (11/26) of these social media websites directly traffic to both LOPs and IOPs, 50% (13/26) of the sites direct traffic to only IOPs and merely 8% (2/26) of the websites direct traffic to only LOPs. Amongst the various social media (Tabular array 4), we found that Facebook directs the highest traffic to both IOPs (58%) and LOPs (42%), far exceeding the second highest (Reddit), which directs xx% traffic to IOPs and 15% to LOPs.

Table 4

Traffic from social media websites to online pharmacies.

Social media	Proportion of traffic to legitimate pharmacies (n=24), %	Proportion of traffic to illicit pharmacies (due north=42), %
Facebook	58	42
Reddit	15	twenty
YouTube	fourteen	xi
Twitter	4	—a
LinkedIn	2	—
Askville	—	7
Pinterest	—	4
Others	7	16

Furthermore, our country data (Table 5) bear witness traffic from 52 countries for the 155 online pharmacies for which nosotros were able to collect country information. Traffic from 27 (52%) countries points to but IOPs, and simply three (six%) countries have traffic simply to LOPs. In addition, the United States is the main consumer for online pharmacies, representing the highest proportion of traffic to both LOPs (97%) and IOPs (60%), among all countries.

Table 5

Traffic from different countries to online pharmacies.

Countries	Proportion of traffic to legitimate online pharmacies (north=thirty), %	Proportion of traffic to illicit online pharmacies (n=139), %
United states	97	71.1
Canada	one	—a
Republic of india	1	6.7
United Kingdom	—	7.6
Others	1	14.6

Tabular array 6 shows the average appointment metrics across the LOPs and IOPs. This shows that the average monthly views (in millions), the number of pages viewed, and time spent on the sites of LOPs are all higher than those of IOPs, whereas the bounce rate (the percentage of visitors leaving the website subsequently viewing only 1 page) is lower for LOPs than that for IOPs. This indicates that when consumers enter the OP websites, they seem to be more engaged with LOPs than IOPs. Withal, there are large variances in the monthly views of the websites, reflecting the huge differences among the websites in IOP likewise equally in LOP. For example, among the licit ones, cvs.com and walgreens.com are definitely the giants, whereas many others but attract a small number of views. In addition, although the average time spent on LOP sites (five min) was significantly college than that of IOPs (three.3 min) with P<.001, the maximum time spent on the sites of IOPs (17.4 min) was much higher than that on LOPs (10.six min). This indicates that when consumers are interested, they may go very engaged with an IOP, leading to potential transactions. Hence, it is imperative to warn the patients before they enter a potential IOP, using the prediction method proposed by the states.

Table half dozen

Consumers' engagement with online pharmacies.

Types of the online pharmacies	monthly views in millions, mean (SD)	Number of page views, mean (SD)	Bounce rate, mean (SD)	Fourth dimension on site in minutes, hateful (SD)
Legitimate online pharmacy	1.48 (3.05)	vii.2 (iii.5)	32.two (16.1)	v.0 (ii.vii)
Illicit online chemist's shop	0.02 (0.05)	4.0 (2.1)	49.4 (17.9)	3.3 (2.two)

OP Status Prediction Models

We now study the performance of the RRPM and RKNN models in their prediction. We consider four performance measures: accuracy, kappa, sensitivity, and specificity. Although sensitivity describes the percentage of IOPs correctly identified, specificity describes the per centum of LOPs correctly identified. Nosotros can run into that Type I fault=ane−specificity and Type 2 mistake=ane−sensitivity. As discussed, we chose the threshold T to pursue a minimum type 2 fault, that is, a maximum sensitivity. In improver, the developed model should have good accurateness and reasonable kappa values, where kappa measures the agreement between observed and predicted classes considering to some extent the possibility of agreement by gamble [xiii].

With 10-fold cross-validation [14], the performance metrics of RKNN (with K=i-9) and RRPM are shown in Table 7. It can be observed that all the RKNN models achieved 100% sensitivity. Nevertheless, the specificity, accuracy, and kappa kickoff increase and and then decrease every bit Chiliad increases with R2NN performing the best, showing excellent metrics. RRPM as well performs reasonably well, achieving a sensitivity of 99.two%, with relatively lower values for kappa and specificity. When irresolute the threshold T for RRPM from the current relatively conservative value to be the reliability score maximizing the model accuracy in the preparation data ready, model accuracy, kappa, and specificity all we improve much. But sensitivity slightly dropped, as expected (Tabular array 7).

Table seven

Performance of the nomenclature models.

Model	Accurateness	Kappa	Specificity	Sensitivity
R1NNa	0.984	0.844	0.76	one
R2NNa	0.986 b	0.859	0.78	1
R3NNa	0.979	0.789	0.68	one
R4NNa	0.975	0.729	0.62	one
R5NNa	0.975	0.729	0.62	1
R6NNa	0.972	0.711	0.58	ane
R7NNa	0.965	0.600	0.46	one
R8NNa	0.954	0.431	0.30	one
R9NNa	0.949	0.321	0.22	one
RRPMc	0.950	0.434	0.36	0.992
RRPM (alternative threshold)	0.968	0.648	0.78	0.977

Implementing RRPM and RKNN on Google Search Results

Our traffic analysis showed that search accounts for the highest traffic to IOPs (39.27%). Our prediction model can be used in a couple of ways for search engines: (i) it tin can exist incorporated on top of search results to filter/flag search results that are likely IOPs and (2) the reliability scores of the OPs can be used to rank the results such that more reliable OPs would appear offset. Therefore, we tested our model on Google search results for 3 popular drugs.

Xanax (alprazolam) is a type of benzodiazepine. More than 30 percent of overdoses involving opioids also involve benzodiazepines [15]. Anecdotal evidence indicates that such drugs are typically the target of IOPs. Nosotros monitored the top keywords that straight traffic to OPs and identified that keywords with the drugs' names contributed to more than traffic than keywords without drug names. Hence, nosotros chose buy Xanax online as the keyword and collected the top 100 search results for the keyword search on September ix, 2016. About all the search results were pharmacies selling Xanax on the web. As a result of the opioid crisis, forth with buy Xanax online, we besides studied the search results of the keywords buy opioids online and buy OxyContin online on April 22, 2017. OxyContin carries a boxed alarm and contains oxycodone, a Schedule Ii controlled substance with an abuse potential similar to other Schedule II opioids.

To test our results, we hand nerveless the status of the OPs obtained through the elevation 100 search results from the NABP and Legitscript. Tabular array 8 provides the status from both sources. Results demonstrate that neither source has an exhaustive database, although Legitscript (which only allows checking 10 pharmacies daily without a fee) has a bigger database, confirming what was constitute by Mackey et al [16]—that hand or automated search of opioid-related sites results in websites non covered past the Legitscript database. Information technology is alarming to note that none of the pharmacies from the top 100 search results are legitimate past definition of either NABP or Legitscript. Nosotros so used RRPM and RKNN to predict the condition of these pharmacies and compared our prediction results with the OP status co-ordinate to Legitscript and NABP (Table 8).

Tabular array eight

Condition of the search results according to Legitscript and National Association Board of Pharmacies.

Keywords searched	IOPa by NABPb	LOPc by NABP	Unknown from NABP	IOP/rogue by Legitscript	LOP/safe by Legitscript	Unknown from Legitscript
Purchase Xanax online	11	0	89	48	0	52
Buy Opioids online	six	0	94	34	0	66
Buy OxyContin online	ten	0	90	25	0	75

Every bit our model relies on the referral data, when the referral data for a particular online pharmacy is not available, its status is defined as unknown by our model. Table 9 compares the prediction results from RRPM and R2NN, respectively, with those from the Legitscript and NABP databases (NABP numbers are shown in parentheses) for the pharmacies obtained from the superlative 100 search results for the 3 keyword searches (hence, 300 overall). For instance, according to Tabular array 9, 104 (27) pharmacies are correctly predicted illicit and 2 (0) are incorrectly predicted equally legitimate pharmacies by RRPM when compared with the status defined by Legitscript (NABP). In improver, the status of 7 (0) IOPs according to the Legitscript (NABP) database cannot be identified by RRPM considering of the lack of referral data.

Table 9

Comparison of the predicted status of online pharmacies based on reference rating prediction method (RRPM) and reference-based K-nearest neighbor (RKNN) with those obtained from Legitscript and National Association Lath of Pharmacies (NABP) databases, with NABP numbers in parentheses.

Prediction results		Status obtained from Legitscript and NABP databases (NABP numbers in parentheses)
		Illicit	Legitimate	Unknown
Status estimated by RRPMa
	Illicit	104 (27)	0 (0)	147 (224)
	Legitimate	2 (0)	0 (0)	3 (5)
	Unknown	7 (0)	0 (0)	37 (44)
Status estimated by R2NNb
	Illicit	106 (27)	0 (0)	145 (225)
	Legitimate	0 (0)	0 (0)	five (5)
	Unknown	7 (0)	0 (0)	37 (43)

Excluding those that are unknown from the corresponding databases, the tables show that RRPM and R2NN produced an mistake charge per unit of 7.96% (0%) and 6.20% (0%), respectively, based on the Legitscript (NABP) database. The above results provide evidence that the proposed prediction models tin can predict online pharmacies with reasonably good accuracy.

Discussion

Comparison With Previous Work

In this study, we conducted a web traffic and engagement analysis of IOPs and LOPs, developed simple prediction models of the status of the OPs based on referral links, and tested the prediction models with information for 763 online pharmacies. Although the previous literature shows evidence of drug selling through IOPs, there has been very express work on the traffic to these websites. One exception is the report by Mackey et al [5], which estimated traffic to an IOP through fictitious advertisements for selling drugs without prescription that they created on social media. In contrast, nosotros collected truthful data on the traffic analysis and the prediction models.

Similarly, very express research is related to identifying and predicting the status of OPs. The study by Fittler et al [10] aimed to place the indicators of IOPs past evaluating 136 of them based on the longevity of the site, geographical location, display of contact data, medical data commutation, prescription requirement, and pharmacy legitimacy verification. They identified that the prescription requirement or availability of contact information does not correlate with illicit pharmacy status as indicated by Legitscript; still, the long-term continuous operation of the website has a stiff correlation with illicit activities. They did not develop a prediction model.

Predicting the status of OPs is related to classifying different websites into certain categories. In general, at that place are two types of approaches: content based and structure based. Hybrid methods also exist. While content-based classification [17] utilizes the website content to allocate the website, structure-based classification exploits the patterns in the link structure or the topology of the hyperlinks of the websites. For example, Amitay et al [18] used structural information to classify 8 classes of websites (eg, corporate sites, search engines, and eastward-store), with the precision of certain classes exceeding 85%. Our prediction model is construction based, and it is like shooting fish in a barrel to see what we attempt to classify as IOP and LOP is much more subtle.

Enquiry on the prediction/classification of LOPs and IOPs is very scarce. The only other work is the report past Corona et al [19] aimed at building a database of OPs using textual content analysis. Note that content-based prediction could be more hands manipulated than structure-based prediction. For instance, if the prediction is based on certain content appearing on the websites, and then IOPs could delete or change the content to confuse the model by making it merely like LOPs. Toward this end, this paper proposes a novel nevertheless simple structure-based idea using relationships among referral websites to predict the condition of OPs.

Finally, when searching the literature for general prediction and nomenclature of websites selling apocryphal products (not limited to drugs), simply 2 studies were found [20,21]. Both used content analysis in general, achieving an accurateness of 86.4% [twenty] and 88% [21]. Our approach tin potentially exist applied to more than full general products than but drugs.

Limitations

Equally nosotros propose a new methodology, nosotros face many limitations. First, because of the limited source of the available ground truth of the status of online pharmacies and the data related to traffic assay, nosotros take a relatively small sample size (for some of the traffic analysis). Nosotros wait that a larger sample size when bachelor will improve the accurateness of the results and let more than detailed assay. When using Google search results, nosotros also confront many websites whose truthful status is unknown; hence, evaluation of our methods using the Google search results presented in our paper is express. Second, the current website information sources (SEMRush and Similarweb) do not provide reliable (or any) information for pocket-size websites lacking sufficient information for traffic overview. Accordingly, the findings of this research are mostly applicable to larger legitimate and illicit web-based players. Third, our proposed method relies on referral website data. Our electric current referral database from the information nosotros collected seemingly works well. However, obviously, the bigger the referral link database, the better. Although outdated links do not hurt the performance (they will non be used), updating these links as more basis truth data becomes available would be desired. Finally, we focus on proposing a novel structure-based prediction framework and developing elementary models to help resolve an important and practical trouble. More advanced models, such every bit a hybrid of structure based and contexture based, can exist developed in the future to farther improve functioning.

Applications and Conclusions

Previous research shows that illicit online pharmacies are present and widely accessed, posting dramatic risks to the drug supply chain integrity and patient health. All the same, because of the sheer scale of this problem (>thirty,000 OPs) and the dynamic nature of online channels, even identifying and monitoring IOPs, the first step to adjourn IOPs, is a difficult task. In this study, we aimed to fill this gap by conducting a traffic assay to increment our agreement of IOPs and proposed a new thought to predict the OP status based on referral data and developed ii specific prediction models (RRPM and RKNN) using this idea. Testing these models on a data set with 763 online pharmacies showed that both models performed well, with an accuracy of 95.0% (RRPM) and 98.6% (RKNN). R2NN outperformed RRPM in more comprehensive metrics (sensitivity, kappa, and specificity). When implementing both models on the Google search results for 3 drugs, we simply incurred an mistake charge per unit of 6.20% for the pharmacies whose true status was known according to the Legitscript database when using the R2NN model and an error rate of 7.96% when using RRPM for the prediction. Although further testing with a larger information prepare is being pursued (the difficulty is the limited footing truth data), we believe our traffic analysis and the approach to employ web analytics of referral websites to predict the status of OPs is among the commencement in the drug field and proposes a feasible and patently constructive fashion to monitor OPs.

The adult framework/models have numerous exciting application areas. For example, they tin exist implemented by search engines, social media, web-based markets (eg, Amazon), and payment companies (eg, Visa and Master cards) to filter IOPs or have the condition of the online pharmacies into consideration when ranking search results, deciding advertising allocations, making payments, disqualifying vendors, or at least warning consumers of potential IOPs. They can also be used with search engines and social media to develop a alert arrangement to assist consumers brand informed decisions. The timeliness of this work could assistance address the current opioid crisis. Policy makers, government agencies, patient advocacy groups, and drug manufacturers may also employ such a organization to identify, monitor, curb IOPs, and brainwash consumers.

Given that this is a critical surface area of business organisation to patients' health and the integrity of the drug supply chain, nosotros promise this study will inspire boosted efficient and effective prediction models or additional applications for the prediction models developed. On a larger scale, nosotros hope to inspire more inquiry in other aspects to fight IOPs. Finally, our literature review also reveals that literature on automatic prediction/identification of websites selling apocryphal products (not limited to drugs) is also very scarce, although selling counterfeit products on the web is a prevalent trouble. Our framework and prediction models can be applied to other products, and we hope to inspire enquiry in this general area likewise.

Abbreviations

IOP	illicit online pharmacies
KNN	Grand-nearest neighbour
LOP	legitimate online pharmacies
NABP	National Association Board of Pharmacies
RKNN	reference-based K-nearest neighbor
RRPM	reference rating prediction method

Appendix

Multimedia Appendix 1

More details of the engagement and traffic source data.

Footnotes

Conflicts of Interest: None declared.

References

2. Quon BS, Firszt R, Eisenberg MJ. A comparison of make-proper noun drug prices betwixt Canadian-based Internet pharmacies and major U.s.a. drug chain pharmacies. Ann Intern Med. 2005 Sep 20;143(6):397–403. doi: x.7326/0003-4819-143-6-200509200-00004. [PubMed] [CrossRef] [Google Scholar]

5. Mackey TK, Aung P, Liang BA. Illicit Internet availability of drugs subject to remember and patient safety consequences. Int J Clin Pharm. 2015 Dec;37(half dozen):1076–85. doi: 10.1007/s11096-015-0154-8. [PubMed] [CrossRef] [Google Scholar]

8. Jena AB, Goldman DP, Foster SE, Califano JA. Prescription medication abuse and illegitimate cyberspace-based pharmacies. Ann Intern Med. 2011 December 20;155(12):848–50. doi: 10.7326/0003-4819-155-12-201112200-00008. [PubMed] [CrossRef] [Google Scholar]

ten. Fittler A, Bősze G, Botz L. Evaluating aspects of online medication safe in long-term follow-up of 136 internet pharmacies: illegal rogue online pharmacies flourish and are long-lived. J Med Net Res. 2013 Sep x;15(nine):e199. doi: 10.2196/jmir.2606. https://www.jmir.org/2013/9/e199/ [PMC gratuitous article] [PubMed] [CrossRef] [Google Scholar]

xi. Gear up E, Hodges JL. Discriminatory assay. Nonparametric discrimination: consistency properties. Int Stat Rev. 1989 Dec;57(iii):238. doi: 10.2307/1403797. [CrossRef] [Google Scholar]

13. Khun M, Johnson Thou. Practical Predictive Modeling. New York, U.s.a.: Springer Science & Business Media; 2013. [Google Scholar]

xiv. Cawley GC, Talbot NL. Efficient leave-one-out cross-validation of kernel fisher discriminant classifiers. Design Recogn. 2003 November;36(eleven):2585–92. doi: 10.1016/s0031-3203(03)00136-v. [CrossRef] [Google Scholar]

sixteen. Mackey T, Kalyanam JJ, Klugman J, Kuzmenko E, Gupta R. Solution to find, classify, and report illicit online marketing and sales of controlled substances via twitter: using machine learning and web forensics to combat digital opioid access. J Med Net Res. 2018 Apr 27;20(four):e10029. doi: 10.2196/10029. https://www.jmir.org/2018/4/e10029/ [PMC complimentary article] [PubMed] [CrossRef] [Google Scholar]

xviii. Amitay Due east, Carmel D, Darlow A, Lempel R, Soffer A. The Connectivity Sonar: Detecting Site Functionality by Structural Patterns. Proceedings of the Fourteenth ACM Briefing on Hypertext and Hypermedia; HYPERTEXT'03; June 11-15, 2003; Maryland, United states of america. 2003. [CrossRef] [Google Scholar]

19. Corona I, Contini M, Ariu D, Giacinto One thousand, Roli F, Lund G, Marinelli Thousand. PharmaGuard: Automatic Identification of Illegal Search-Indexed Online Pharmacies. 2nd International Conference on Cybernetics; CYBCONF'15; June 24-26, 2015; Gdynia, Poland. 2015. [CrossRef] [Google Scholar]

20. Wadleigh J, Drew J, Moore T. The E-Commerce Market for 'Lemons': Identification and Assay of Websites Selling Apocryphal Goods. Proceedings of the 24th International Conference on World wide web; World wide web'fifteen; May xviii-22, 2015; Florence, Italy. 2015. [CrossRef] [Google Scholar]

21. Carpineto C, Romano Thousand. Learning to Detect and Measure Imitation Ecommerce Websites in Search-engine Results. Proceedings of the International Briefing on Web Intelligence; WI'17; September 25-29, 2017; Leipzig, Deutschland. 2017. [CrossRef] [Google Scholar]

finleywhid1985.blogspot.com

Source: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7479587/

Share this post