Category Archives: Librarian Lens

Librarian Lens, Uncategorized

Librarian Lens: “All of Us” Research Program

Leave a comment

The Librarian Lens is an occasional column featuring librarians who support the research lifecycle across a range of disciplines sharing research tips, updates about both Libraries-provided and open source resources, and related topics intended to intrigue, demystify and highlight topics of interest to the research-curious. Posted columns are provided or curated by librarians from the University of Texas Libraries STEM and Social Science Engagement Team.

Calling all researchers, students and anyone interested in exciting developments in health and medical information…read on to find out about the exciting All of Us Research Program that is on a mission to accelerate medical breakthroughs!

The All of Us Research Program is part of the National Institutes of Health. It is actively collecting data from a diverse population of participants across the United States. As of April 1, 2024 the program has 783,000+ participants, 432,000+ electronic health records and 555,000+ biosamples received according to the program’s Data Snapshots, which are updated daily. The program continues to enroll people from all backgrounds and is aiming for 1,000,000+ participants.

The Data

The program aims to connect the research community with the data, which is longitudinal and from a diverse population, including those typically underrepresented in biomedical research. The data are from participants’ survey responses, electronic health records, physical measurements, Fitbit records and biosamples. DNA from participants’ biosamples supply genomic data, including genotyping arrays and whole genome sequences. The data undergoes a curation process with rigorous privacy protections. On average the data is pushed out once a year. The All of Us Research Hub stores the data in a secure Cloud-based space, where it must remain. Imagine the possibility of all the research project opportunities, especially those on precision medicine, with the ability to combine genomic data with phenotypic data and analyze it on a large scale!

Public Access

There is a terrific way to get a better idea of the data that are available whether out of curiosity or for more in-depth study and that is through the program’s Data Snapshots and interactive Data Browser that anyone can access through the All of Us Research Hub. No prior authorization or account is required to explore the aggregated, anonymized data. In fact, this public access data can serve as a great jumping off point for developing a research project question! In addition, the Research Projects Directory, which currently  lists over 10,000 projects, can be viewed by anyone and is a great place to browse for inspiration or collaboration opportunities. For students, it is a wonderful way to see examples of scientific questions and approaches. Papers resulting from these research projects are listed on All of Us Publications and anyone can read them because they have been published open access. The total number of publications is close to 300!

The DNA double helix rests on a field of ACGTs and binary numbers.
Credit: Jonathan Bailey, NHGRI. Public Domain

Researcher Access

The All of Us Research Hub provides access to more granular data along with tools for analysis through the Researcher Workbench. To access granular data either on the registered tier or the controlled tier, researchers must complete registration and be approved. Registration is a four step process. The first is to confirm your institution has a Data Use and Registration Agreement (DURA) in place with All of Us. At the time of this writing, the University of Texas at Austin has a DURA in place. The next step is to create an account and verify your identity. The third is to complete mandatory training. Additional training is required to analyze the controlled tier data because that is where the genomic data is stored. The fourth and final step is to sign the Data User Code of Conduct. Steps two through four take about two hours to complete but do not have to be done all in one sitting. After completing registration, researchers will be able to access the data once they receive notification of approval from the All of Us program. The data can then be accessed and analyzed through the Researcher Workbench where researchers create Workspaces and submit descriptions of their research projects for listing on the Research Projects Directory. The program makes it easy for researchers to invite other registered users from their own institutions or from other institutions to their Workspaces for collaboration. This is especially helpful if a researcher needs someone else to run analysis with R or Python, the two options currently available. The program encourages and supports team science!

Access Cost & User Support

Even though there is a cost to researchers for analyzing and storing data for their research projects in the Cloud, there is good news. The program provides researchers $300 initial credits upon creating Workspaces in the Researcher Workbench to help them get started on their research projects. Researchers can get information on how to run their data cost effectively from the program’s User Support Hub. They can also speak with the program’s support team to get an idea of cost before analyzing data.

The All of Us Research Program’s investment in medical breakthroughs could not be more clear with the amount of thoughtful, varied support it has put together to help researchers access and analyze the data. The User Support Hub has video tutorials, articles, a help button, and an event calendar that lists the support team’s upcoming office hours among other events. Support resources for the Researcher Workbench range from getting started to working with the data to credits and billing. The support articles are kept current and are on topics such as using RStudio on the Researcher Workbench. The support team is responsive and ready to help! Help is provided by actual people, not chatbots. The program welcomes feedback and uses it to make the program better. For instance, if there is certain data not being collected that would be helpful to collect, let the program know!

iPad Showing Genomic Data
Credit: Darryl Leja, NHGRI. Public Domain

The Impact

By collecting data from the United States’ diverse population, especially from traditionally underrepresented groups in biomedical research, the All of Us Research Program strives to make the medical data that is currently available to researchers more complete. With researchers having access to not only genomic data, but also phenotypic data, their studies can fuel insights into our health on an individual level; thus potentially allowing for healthcare providers to provide better, more tailored care to each of us. To see the research studies made possible so far by the availability of the All of Us data, take a look at the All of Us Research Highlights today, and check back regularly to keep an eye on advancements and breakthroughs!

The All of Us Research Program seeks to connect the research community with its dataset, which is one of the largest, most diverse ever assembled, to spur research that will improve health for all of us. If you are feeling inspired as a researcher or want to become a participant or are just curious to find out more about the program, visit the program’s websites at researchallofus.org or allofus.nih.gov.

Librarian Lens

Librarian Lens: Customer Reviews Data

Leave a comment

The Librarian Lens is an occasional column featuring librarians who support the research lifecycle across a range of disciplines sharing research tips, updates about both Libraries-provided and open source resources, and related topics intended to intrigue, demystify and highlight topics of interest to the research-curious. Posted columns are provided or curated by librarians from the University of Texas Libraries STEM and Social Science Engagement Team.

Customer reviews data are what you look at when you’re deciding which product to buy, which restaurant to eat at, or which hotel to reserve. The data consists of star ratings and the written or video reviews from customers. Most people consult customer review data when making online purchasing decisions.

Customer reviews are also considered a form of advertising. In advertising lingo, customer review data is known as “electronic word of mouth,” meaning it comes from a customer’s experience, not the manufacturer or service provider. Traditional “word of mouth” advertising, especially from people we know, is the most trusted form of advertising. Electronic word of mouth can also influence purchase behavior. Academic researchers have measured that influence and concluded that the most trustable and influential reviews are those that are high quality, i.e., they contain a lot of detail and the reviewer posts regularly.

However, is trusting reviews from people we don’t know a good idea? Like the answer to many questions involving human-centered behaviors, the answer is, it depends.

On its face, customer review data is a compilation of crowd sourced wisdom. If we believe that people are genuinely reporting their experience with the product or service, then why not rely on customer review data?

Guilt

There are many reasons why people don’t always give their honest opinions in a review. They may feel guilty about leaving a poor review. This is common for services such as ridesharing or house rentals in which a consumer’s written opinion can impact the service provider’s ability to earn income. It also happens because service providers can review the consumer. No one wants to be labeled as difficult. These reviews are not always reliable.

Distraction/Hidden Agenda

Other reasons for consumers not giving honest reviews may be unrelated to their experience with the product or service. They could be hungry or in a hurry or they may not read the product’s instructions. Or they could have a political disagreement with the company. There is no way for you as a consumer to know about these conditions unless you research the reviewer and check their other reviews. You may not have time for this kind of investigation.

Consumer Reports tests electric lights in different ways, including for longevity and brightness. Consumer Reports, CC BY-SA 4.0 https://creativecommons.org/licenses/by-sa/4.0, via Wikimedia Commons.

Expertise

Another reason to consider not using customer review data is because people leaving the reviews may not be experts in the field. In fact, most reviews are from people who do not have special expertise. You may have noticed this in movie reviews. Film critics review films differently from most people, such as in the case the Netflix show Squid Game: The Challenge. This is why it is interesting to look at a site like Rotten Tomatoes which contains reviews from both experts (Tomatometer) and regular folks (Audience Score).

In situations in which you are not spending a large amount of money, it may not matter all that much what the reviewers write. After all, how much difference is there between one type of hand lotion and another? More details help. The reviewer could write I live in a dry environment, and this product improved my skin’s texture. This why sites that sell clothing often ask reviewers to fill in other criteria such as age and body type so that you can try to choose a reviewer that matches you so that you can use their review to make your best guess about whether to make a purchase.

If you are spending a large amount of money, or procuring something for a child, it’s a good idea to use expert reviews. For investment advice or car purchases, please turn to the folks who work in this field. Or for safety considerations for items such as car seats or booster seats, consult the National Highway Traffic Safety Administration. For technical products, such as hardware or networking equipment, use sources such a CNET or PC Magazine.

Another consideration is durability. Some organizations, such as Consumer Reports and Wirecutter are testing experts. They don’t specialize in a particular industry, but they have expertise in testing. For home products such as microwaves or washing machines, it’s a good idea to read reviews from these organizations, because they’ve done the kind of tests where they’ve held a watch under water for 8 hours, or slammed an oven door repeatedly, unlike the regular consumers who post reviews to Best Buy or Lowes. 

Fake Reviews

In a recently published article, New York Times reporter Stuart A. Thompson said that fake reviews are so pervasive that nearly every online shopper has most likely encountered one.  Amazon blocked more than 200 million suspected fake reviews last year and Google said it removed 115 million rule-breaking reviews from Maps in 2022. It is an on-going problem. In some case, people are paid to write fake reviews, so the reviews are coming from individuals, not bots. Sometimes fake review writers are given a script to use, so searching for a phrase that comes up repeatedly could indicate a fake review.

Artificial Intelligence

Artificial Intelligence, or AI could help or hinder the fake review problem. You may have noticed that Amazon now provides summaries of reviews for products. If you click on Reviews, there’s a section that says “Customers say” which summarizes comments about product attributes. In small text beneath the summary paragraph, it says “AI generated from text of customer reviews.” It’s likely that fake reviews are included in this collection, and that skews the summary to be more positive, since fake reviews are almost always positive. One way to combat this is to read the negative reviews.

Customer Review Data was the subject of a UT Libraries’ Data & Donuts Workshop in October 2023. Please link to the Zoom recording for more information on customer reviews.