Data sharing and privacy issues in neuroimaging research: Opportunities, obstacles, challenges, and monsters under the bed

May combine two claims.

In the authors' longitudinal study of child development, children were given several photos of their brains following their session, and some of the children placed these on social media.

“Finally, in our longitudinal study of child development (White et al., 2017), we give children several photos of their brains (i.e., sagittal midline slice from the structural MRI) following their session, of which we have learned that some of the children have placed on social media.”

• cites White et al., 2017Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The GDPR's 'Privacy by Design' requires that the safeguarding of data should be discussed and implemented during the design phase of the study.

“In addition, the GDPR's “Privacy by Design” requires that the safeguarding of data should be discussed and implemented during the design phase of the study.”

• cites GDPR, 2016Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The GDPR places greater responsibility on institutions to safeguard the privacy of personal data, such as assuring that there is a data controller to monitor data security.

“The GDPR places greater responsibility on institutions to safeguard the privacy of personal data, such as assuring that there is a data controller to monitor data security.”

• cites GDPR, 2016Not checked yet

  Not checked in the free run.

May combine two claims.

Larger sample sizes offer the opportunity to apply more sophisticated statistical models to the data, since most biological processes have non-linear and stochastic mechanisms.

“Larger sample sizes will offer the opportunity to apply more sophisticated statistical models to the data (Bzdok & Yeo, 2017), since most biological processes have non‐linear and stochastic mechanisms (White, 2019).”

• cites Bzdok & Yeo, 2017Not checked yet

  Not checked in the free run.

• cites White, 2019Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The GDPR went into effect on May 25, 2018 and involves data privacy laws involving the storage, transfer, and sharing of data, both within and outside the EU and extending to the European Economic Area.

“The GDPR went into effect on May 25, 2018 and involves data privacy laws involving the storage, transfer, and sharing of data, both within and outside the EU and extending to the European Economic Area.”

• cites GDPR, 2016Not checked yet

  Not checked in the free run.

Classified as probably not a substantial claim.

Within the GDPR a clear distinction is made between personal data, de-identified data, and fully anonymized data.

“Within the GDPR a clear distinction is made between personal data, de‐identified data, and fully anonymized data.”

• cites GDPR, 2016Not checked yet

  Not checked in the free run.

May combine two claims.

Data sharing initiatives open the door for reproducibility, replication, and increased transparency.

“In addition, data sharing initiatives open the door for reproducibility, replication, and increased transparency (Milham et al., 2018).”

• cites Milham et al., 2018Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Meta-analysis does not require advanced analysis methods to account for clustering-effects within cohorts.

“Moreover, it does not require advanced analysis methods to account for clustering‐effects within cohorts.”

May combine two claims.

Comparing demographic and clinical information of those willing and not willing to share provides an indication of representativeness between those groups, but it does not account for representativeness as a result of potential biases during the inclusion phase of the study.

“While this will provide an indication of representativeness between those willing and not willing to share their data, it does not account for representativeness as a result of potential biases during the inclusion phase of the study (i.e., selection bias).”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The combination of neuroimaging and genetics, coined 'imaging genetics' has also emerged, with the Enhancing Neuro Imaging Genetics through Meta Analyses (ENIGMA) consortium playing a leading role in this initiative.

“The combination of neuroimaging and genetics, coined “imaging genetics” has also emerged, with the Enhancing Neuro Imaging Genetics through Meta Analyses (ENIGMA) consortium playing a leading role in this initiative (Thompson et al., 2014).”

• cites Thompson et al., 2014Not checked yet

  Not checked in the free run.

May combine two claims.

A consent form written that precludes the option for data sharing dramatically limits the ability for sharing to occur, although if it is possible to fully anonymize the data, then sharing is possible in most situations.

“A consent form written that precludes the option for data sharing dramatically limits the ability for sharing to occur, although if it's possible to fully anonymize the data, then sharing is possible in most situations.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Under the recent Horizon 2020 call, as far as possible, research data should be made available to 'access, mine, exploit, reproduce, and disseminate (free of charge)' research data.

“Under the recent Horizon 2020 call, as far as possible, research data should be made available to “access, mine, exploit, reproduce, and disseminate (free of charge)” research data (European‐Commission, 2017).”

• cites European‐Commission, 2017Not checked yet

  Not checked in the free run.

May combine two claims.

Sharing unthresholded or network maps involves retrospective storage of completed studies and does not allow novel subject-level models to be run on time series data.

“In addition, these approaches involve retrospective storage of completed studies and do not allow for novel subject‐level models to be run on the time series data.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Performing analyses at a centralized site has the benefit of being able to work within local or regional restrictions, whether defined by law or by the facility.

“Moreover, performing analyses at a centralized site has the benefit of being able to work within local or regional restrictions, whether defined by law or by the facility.”

May combine two claims.

Examples of re-identification, such as highly accurate identification via facial reconstruction and machine learning identification from generative models, challenge the technical and legal adequacy of the de-identification release-and-forget model, spurring calls for additional privacy guidance.

“Examples of re‐identification, such as highly accurate identification via facial reconstruction (Schwarz et al., 2019) and machine learning identification from generative models (Rocher, Hendrickx, & de Montjoye, 2019) challenge the technical and legal adequacy of the de‐identification release‐and‐forget model, spurring calls for additional privacy guidance (Morris, 2019).”

• cites Schwarz et al., 2019Not checked yet

  Not checked in the free run.

• cites Rocher, Hendrickx, & de Montjoye, 2019Not checked yet

  Not checked in the free run.

• cites Morris, 2019Not checked yet

  Not checked in the free run.

May combine two claims.

Multivariate and other advanced analytic approaches can extract a remarkable amount of information from highly distilled features; intrinsic networks can be captured from covariation among individual datasets.

“Multivariate and other advanced analytic approaches have the ability to extract a remarkable amount of information from these highly distilled features, for example, intrinsic networks can be captured from covariation among individual datasets (Calhoun & Allen, 2013; Smith et al., 2009).”

• cites Calhoun & Allen, 2013Not checked yet

  Not checked in the free run.

• cites Smith et al., 2009Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing).

The UK Biobank has the goal to release neuroimaging data from 100,000 participants.

“from the United Kingdom the UK Biobank has the goal to release neuroimaging data from 100,000 participants (Miller et al., 2016)”

• cites Miller et al., 2016Not checked yet

  Not checked in the free run.

May combine two claims.

Pooling smaller datasets expands opportunities to address specific questions and to assess replicability of findings.

“While certain analyses cannot be applied to small datasets, pooling these smaller datasets expands the opportunities to address specific questions and to assess the replicability of the findings.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The perspective of the NIH for nearly two decades has been that 'all data should be considered for data sharing'.

“For example, the perspective of the NIH for nearly two decades has been that “all data should be considered for data sharing” (NIH, 2003).”

• cites NIH, 2003Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

While it may be possible for researchers to design their study and inclusion criteria to include only those participants who are willing to share their data, for clinical studies, this may result in a selection bias.

“While it may be possible for researchers to design their study and inclusion criteria to include only those participants who are willing to share their data, for clinical studies, this may result in a selection bias.”

May combine two claims.

The NIMH currently mandates data sharing (with an institutionally signed DUA) for almost all funded studies, with some notable exceptions for extremely sensitive data which might lead to the ability to re-identify an individual.

“The NIMH currently mandates data sharing (with an institutionally signed DUA (Miller et al., 2017)) for almost all funded studies (there are some notable exceptions to this in the case of extremely sensitive data which might lead to the ability to re‐identify an individual).”

• cites Miller et al., 2017Not checked yet

  Not checked in the free run.

May combine two claims.

While there is support that individuals have characteristic patterns of functional brain connectivity, known as functional connectome fingerprinting, these have not been used to identify individuals.

“While there is support that individuals have characteristic patterns of functional brain connectivity, known as functional connectome fingerprinting, these have not been used to identify individuals (Finn et al., 2015).”

• cites Finn et al., 2015Not checked yet

  Not checked in the free run.

May combine two claims.

In spite of the obstacles, many forms of data sharing are taking place and the benefits of these efforts have been seen.

“Nevertheless, in spite of the obstacles, many forms of data sharing are taking place and the benefits of these efforts have been seen (Milham et al., 2018).”

• cites Milham et al., 2018Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

With the proper safeguards and data agreements, cloud-based computing will be equally as safe as an encrypted hard drive with a strong password that is behind locked doors.

“With the proper safeguards and data agreements, cloud‐based computing will be equally as safe as an encrypted hard drive with a strong password that is behind locked doors.”

May combine two claims.

The reason why data sharing initiatives are gaining momentum is because they foster collaboration and can advance the pace of scientific discovery.

“The reason why these initiatives are gaining momentum is because they foster collaboration and can advance the pace of scientific discovery.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

If participants provide consent for their data to be shared, then they can enjoy knowing that researchers across the globe are potentially working with their data to better understand the complexities of brain structure and function and to bring about novel discoveries.

“Finally, if participants provide consent for their data to be shared, which is the most important element, then they can enjoy knowing that researchers across the globe are potentially working with their data to better understand the complexities of brain structure and function and to bring about novel discoveries.”

Probably not a substantial claim: may not pass S4 (based).

Preparing data for sharing is work above and beyond what is needed for a group to analyze data locally.

“Preparing data for sharing is work above and beyond what is needed for a group to analyze data locally.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Researchers are evaluated not on data sharing, but rather based on publications and grants, thus there is less incentive to engage in the considerable effort necessary to make a dataset available for sharing.

“Researchers are evaluated not on data sharing, but rather based on publications and grants, thus there is less incentive to engage in the considerable effort necessary to make a dataset available for sharing.”

• AssumptionRests on an assumption: Research careers are primarily evaluated on the number, quality, and impact of papers published and on the acquisition of grant funding, which depends on the number, quality and impact of publications.
• AssumptionRests on an assumption: Data collection takes considerable time and effort to assure that the quality of the data is high and the data is properly cleaned and used appropriately.

May combine two claims.

Including metrics that allow users to understand underlying quality of pre-processed images reduces noise-related variability and increases power.

“Assuring that the shared data has the optimum data quality, or including metrics that allow users to understand the underlying quality of pre‐processed images (Esteban et al., 2017; White et al., 2018) is important to reduce noise‐related variability and to increase power (Zuo, Xu, & Milham, 2019).”

• cites Esteban et al., 2017Not checked yet

  Not checked in the free run.

• cites White et al., 2018Not checked yet

  Not checked in the free run.

• cites Zuo, Xu, & Milham, 2019Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

For those living in the United States, most human subjects research falls under the Common Rule, which is based on the 1975 revision of the Declaration of Helsinki.

“For those living in the United States, most human subjects research falls under the “Common Rule” (45 C.F.R. § 46 Subpart A), which is based on the 1975 revision of the Declaration of Helsinki.”

• cites 45 C.F.R. § 46 Subpart ANot checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Data sharing requires that those who use the data have a good understanding of the key variables, including the nuances of the data.

“Data sharing requires that those who use the data have a good understanding of the key variables, including the nuances of the data.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Data can be shared with investigators from low- and middle-income countries who may not have the resources to conduct expensive neuroimaging studies, but do have the ability to ask interesting and creative questions of the data.

“In addition, data can be shared with investigators from low‐ and middle‐income countries who may not have the resources to conduct expensive neuroimaging studies, but do have the ability to ask interesting and creative questions of the data.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Under some regulations and certain countries, data would need to be fully anonymized prior to being uploaded to OpenNEURO.

“Thus, under some regulations and certain countries, the data would need to be fully anonymized prior to being uploaded.”

• AssumptionRests on an assumption: Sharing data within OpenNEURO has the requirement that, following a 36-month grace period following the first successful analysis of the data, the data will become publicly available under a Creative Commons (CCO) license.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

In light of recent legal changes, notably in the EU, ethical aspects related to data have actually become more straightforward.

“In light of recent legal changes, notably in the EU, ethical aspects related to data have actually become more straightforward.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Specific state regulations such as the Federal Policy for the Protection of Human Subjects or the Common Rule still apply.

“However, specific state regulations such as the “Federal Policy for the Protection of Human Subjects” or the Common Rule still apply.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Providing some time for those who collected the data to write up results can typically circumvent the issue of others publishing sooner.

“Providing some time for those who collected the data to write up the results, however, can typically circumvent this issue.”

• AssumptionRests on an assumption: The most common issue making researchers less willing to share is that collecting and collating data takes considerable effort and others could publish results sooner than those who collected the data.

Classified as probably not a substantial claim. May combine two claims.

HIPAA applies to covered entities only; covered entities are health plans, health care clearinghouses, and health care providers electronically transmitting health information in connection with transactions for which HHS has adopted standards.

“HIPAA applies to covered entities only, covered entities are health plans, health care clearinghouses, and health care providers electronically transmitting health information in connection with transactions for which Health and Human Services (HHS) has adopted standards (45 C.F.R. § 160.103).”

• cites 45 C.F.R. § 160.103Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

The most notable difference between the GDPR and HIPAA is to whom the regulations apply.

“The most notable difference between the GDPR and HIPAA is to whom the regulations apply.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

In some cases, for example, rare disease, or an extremely high-profile scientific focus, the risk level may be too high for sharing.

“In some cases, for example, rare disease, or an extremely high‐profile scientific focus, the risk level may be too high for sharing.”

May combine two claims.

If certain associations reach sub-threshold significance at individual sites, they will not be taken into account in pooled meta-analysis, whereas in mega-analysis, those associations might be discovered because of increased power.

“If certain associations reach sub‐threshold significance at individual sites, they will not be taken into account in the pooled meta‐analysis, whereas in a mega‐analysis, those associations might be discovered, simply because of the increased power.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Research involving PHI from non-covered institutions does not fall under HIPAA, but rather under the Common Rule.

“Research involving PHI from non‐covered institutions does not fall under HIPAA, but rather under the Common Rule.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Pure reproduction can only be established with detailed information on the coding of variables, the approach to missing data, and how the analyses were performed.

“Pure reproduction can only be established with detailed information on the coding of variables, the approach to missing data, and how the analyses were performed.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

While the large number of authors may dilute the impact of those authors contributing the most work, without rewarding via authorship would likely impair initiatives such as the ENIGMA consortium.

“While the large number of authors may dilute the impact of those authors contributing the most work, without rewarding via authorship would likely impair initiatives such as the ENIGMA consortium.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Research studies involving neuroimaging require approval from the local medical ethics committee or institutional review board in accordance with the Declaration of Helsinki.

“Research studies involving neuroimaging require approval from the local medical ethics committee or institutional review board in accordance with the Declaration of Helsinki (WMA, 2000).”

• cites WMA, 2000Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

As certain institutions have both covered and non-covered functions, there is a possibility to elect for being a hybrid entity, where only the covered functions must comply with the HIPAA requirements under the Privacy Rule.

“As certain institutions have both covered and non‐covered functions, there is a possibility to elect for being a hybrid entity, where only the covered functions must comply with the HIPAA requirements under the Privacy Rule.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

Attorneys whose job is to protect the university may side on being overly risk aversive, limiting the risk to a university, while at the same time potentially limiting the advancement of scientific discovery.

“Attorneys whose job is to protect the university may side on the being overly risk aversive, limiting the risk to a university, while at the same time potentially limiting the advancement of scientific discovery.”

Probably not a substantial claim: may not pass S3 (load-bearing).

PHI not held by a covered entity can be used and disclosed without regard to the Privacy Rule.

“PHI not held by a covered entity can be used and disclosed without regard to the Privacy Rule.”

May combine two claims.

Utilization of standardized approaches can dramatically reduce the time necessary to understand the nature of the data and to reduce the number of errors due to misunderstandings surrounding the data.

“The utilization of standardized approaches can dramatically reduce the time necessary to understand the nature of the data and to reduce the number of errors due to misunderstandings surrounding the data.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Research taking place with personal health information from covered institutions falls under HIPAA.

“However, research taking place with personal health information from covered institutions (i.e., hospitals, clinics, etc.) falls under the HIPAA.”

May combine two claims.

Data sharing allows for greater transparency in science with the ability to promote reproducibility and replication of study findings.

“Data sharing allows for greater transparency in science with the ability to promote reproducibility and replication of study findings (Button et al., 2013; Ioannidis, 2005; Open Science Collaboration, 2015).”

• cites Button et al., 2013Not checked yet

  Not checked in the free run.

• cites Ioannidis, 2005Not checked yet

  Not checked in the free run.

• cites Open Science Collaboration, 2015Not checked yet

  Not checked in the free run.

May combine two claims.

COINSTAC offers fully decentralized (and potentially privatized analysis), allowing the data to remain local at the site of collection, by leveraging local compute resources for each site's data.

“The collaborative informatics and neuroimaging suite toolkit for anonymous computation (COINSTAC [Plis et al., 2016]; https://github.com/trendscenter/coinstac) tool and approach goes a step further in offering fully decentralized (and potentially privatized analysis), allowing the data to remain local at the site of collection, by leveraging local compute resources for each site's data.”

• cites Plis et al., 2016Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S4 (based).

If the data is released relatively quickly after collection, other groups may access the data and publish prior to those who actually have collected the data.

“If the data is released relatively quickly after collection, other groups may access the data and publish prior to those who actually have collected the data.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

There are a number of valid concerns raised by researchers related to data sharing, however, for every concern there is a feasible solution.

“There are a number of valid concerns raised by researchers related to data sharing, however, for every concern there is a feasible solution (Figure 1):”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

These laws, in essence, give much of the control back to the participants.

“These laws, in essence, give much of the control back to the participants.”

May combine two claims.

Specific rules apply for data sharing with countries with similar privacy protection under the GDPR adequacy decision, and thus to share data with other countries, additional safeguards are often necessary.

“Specific rules apply for data sharing with countries with similar privacy protection, (i.e., countries that fall under the GDPR adequacy decision [Council of the European Union and European Parliament, 2016]), and thus to share data with other countries, additional safeguards are often necessary.”

• cites Council of the European Union and European Parliament, 2016Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable).

While there are obstacles to data sharing, most of these can be overcome.

“While there are obstacles to data sharing, most of these can be overcome.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Junior investigators are especially at risk, as they typically need more time to analyze and write the papers.

“Junior investigators are especially at risk, as they typically need more time to analyze and write the papers.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Changing some of the obstacles would likely require changes in policies of funding agencies and journals to provide support and credit for those who make the time and effort to participate in data sharing.

“Changing some of the obstacles would likely require changes in policies of funding agencies and journals to provide support and credit for those who make the time and effort to participate in data sharing.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Clinical neuroimaging data is held by covered entities and HIPAA applies to this data.

“For example, clinical neuroimaging data is held by covered entities and HIPAA applies to this data.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

If a different group publishes first, it may make it difficult for the PhD student to publish, and thus could hamper their career.

“If a different group publishes first, it may make it difficult for the PhD student to publish, and thus could hamper their career.”

Probably not a substantial claim: may not pass S3 (load-bearing).

ViPAR leverages federated databases to provide temporal pooling of the actual data for analysis.

“another approach called ViPAR (virtual pooling and analysis of research data) leverages federated databases to provide temporal pooling of the actual data for analysis (Carter et al., 2016).”

• cites Carter et al., 2016Not checked yet

  Not checked in the free run.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Researchers can provide the opportunity for participants to decide whether they are willing, or would like, that their data be shared.

“Researchers can provide the opportunity for the participants to decide whether they are willing, or would like that their data be shared.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

There is a certain fear that has emerged regarding data sharing where it has become greater than life, monsters under the bed.

“While we do not intend to minimize the importance of data security, there is a certain fear that has emerged regarding data sharing where it has become greater than life, monsters under the bed.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

There are both real and imagined obstacles for data sharing, which equate with not all researchers being supportive of data sharing initiatives.

“There are both real and imagined obstacles for data sharing which equate with not all researchers being supportive of data sharing initiatives.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

The goal of university attorneys may result in a very restrictive bar for individual researchers within certain institutions.

“This may result in the setting of a very restrictive bar for individual researchers within certain institutions.”

• AssumptionRests on an assumption: Attorneys who work for specific universities have the goal to assure that the university is protected from potential legal actions, such as the potential 20 million euro fine imposed by non-compliance with the GDPR.

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

One key element that is often not discussed in regards to data sharing is the wishes of the participant in allowing their data to be shared.

“However, one key element that is often not discussed in regards to data sharing is the wishes of the participant in allowing their data to be shared.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Within the EU and likely other countries, the complexities of laws and fears of retribution may serve as a rationale for some researchers for not sharing.

“Within the EU and likely other countries, the complexities of these laws and the fears of retribution may serve as a rationale for some researchers for not sharing.”

• AssumptionRests on an assumption: The goal of the GDPR is to provide greater control and protection to the individual over whose data has been collected.

Probably not a substantial claim: may not pass S2 (checkable).

There is a balance between protecting individual data and allowing some level of risk/benefit for data sharing.

“Thus, there is a balance.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based). May combine two claims.

Keeping data under lock and key for use by only a handful of researchers may protect privacy but will limit scientific discovery.

“However, there should be a balance, as keeping data under lock and key for use by only a handful of researchers may protect privacy, but will limit scientific discovery.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

Collaborative networks and data sharing initiatives are broadening the opportunities for the advancement of science and the ability to ask important research questions that could benefit others.

“Collaborative networks and data sharing initiatives are broadening the opportunities for the advancement of science and the ability to ask important research questions that could benefit others.”

Probably not a substantial claim: may not pass S4 (based).

For researchers who are not keen on data sharing, recent and emerging regulations regarding human subjects data can be used as a barrier, or excuse, for not taking part in data sharing initiatives.

“For researchers who are not keen on data sharing, recent and emerging regulations regarding human subjects data can be used as a barrier, or excuse, for not taking part in data sharing initiatives.”

Probably not a substantial claim: may not pass S4 (based).

Sharing everything with everyone does not safeguard individual privacy.

“Alternatively, sharing everything with everyone does not safeguard individual privacy.”

Classified as probably not a substantial claim.

Data sharing will entail some level of risk.

“There is no question that data sharing will entail some level of risk.”

Probably not a substantial claim: may not pass S4 (based).

A data leak of sensitive information could result in individuals' data being used for unintended and potentially harmful purposes.

“A data leak of sensitive information, for example, could result in individual's data being used for unintended and potentially harmful purposes.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

If all researchers kept their data under tight control with no data sharing outside their research group, this would hamper the progress of scientific discovery.

“However, if all researchers kept their data under tight control with no data sharing outside their research group, this will hamper the progress of scientific discovery.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Differences in national restrictions on data become equalized to some extent if participants are allowed to decide how their data are used.

“Different countries may differ slightly in the level of restrictions towards data, however, these differences become equalized to some extent if the participants are allowed to decide how they want their data to be used.”