Monday, December 11, 2023

Using machine learning to monitor driver ‘workload’ could help improve road safety

UNIVERSITY OF CAMBRIDGE

Researchers have developed an adaptable algorithm that could improve road safety by predicting when drivers are able to safely interact with in-vehicle systems or receive messages, such as traffic alerts, incoming calls or driving directions.

The researchers, from the University of Cambridge, working in partnership with Jaguar Land Rover (JLR) used a combination of on-road experiments and machine learning as well as Bayesian filtering techniques to reliably and continuously measure driver ‘workload’. Driving in an unfamiliar area may translate to a high workload, while a daily commute may mean a lower workload.

The resulting algorithm is highly adaptable and can respond in near real-time to changes the driver’s behaviour and status, road conditions, road type, or driver characteristics.

This information could then be incorporated into in-vehicle systems such as infotainment and navigation, displays, advanced driver assistance systems (ADAS) and others. Any driver vehicle interaction can be then customised to prioritise safety and enhance the user experience, delivering adaptive human machine interactions. For example, drivers are only alerted at times of low workload, so that the driver can keep their full concentration on the road in more stressful driving scenarios. The results are reported in the journal IEEE Transactions on Intelligent Vehicles.

“More and more data is made available to drivers all the time. However, with increasing levels of driver demand, this can be a major risk factor for road safety,” said co-first author Dr Bashar Ahmad from Cambridge’s Department of Engineering. “There is a lot of information that a vehicle can make available to the driver, but it’s not safe or practical to do so unless you know the status of the driver.”

A driver’s status – or workload – can change frequently. Driving in a new area, in heavy traffic or in poor road conditions, for example, is usually more demanding than a daily commute.

“If you’re in a demanding driving situation, that would be a bad time for a message to pop up on a screen or a heads-up display,” said Ahmad. “The issue for car manufacturers is how to measure how occupied the driver is, and instigate interactions or issue messages or prompts only when the driver is happy to receive them.”

There are algorithms for measuring the levels of driver demand using eye gaze trackers and biometric data from heart rate monitors, but the Cambridge researchers wanted to develop an approach that could do the same thing using information that’s available in any car, specifically driving performance signals such as steering, acceleration and braking data. It should also be able consume and fuse different unsynchronised data streams that have different update rates, including from biometric sensors if available.

To measure driver workload, the researchers first developed a modified version of the Peripheral Detection Task to collect, in an automated way, subjective workload information during driving. For the experiment, a phone showing a route on a navigation app was mounted to the car’s central air vent, next to a small LED ring light that would blink at regular intervals. Participants all followed the same route through a mix of rural, urban and main roads. They were asked to push a finger-worn button whenever the LED light lit up in red and the driver perceived they were in a low workload scenario.

Video analysis of the experiment, paired with the data from the buttons, allowed the researchers to identify high workload situations, such as busy junctions or a vehicle in front or behind the driver behaving unusually.

The on-road data was then used to develop and validate a supervised machine learning framework to profile drivers based on the average workload they experience, and an adaptable Bayesian filtering approach for sequentially estimating, in real-time, the driver’s instantaneous workload, using several driving performance signals including steering and braking. The framework combines macro and micro measures of workload where the former is the driver’s average workload profile and the latter is the instantaneous one.

“For most machine learning applications like this, you would have to train it on a particular driver, but we’ve been able to adapt the models on the go using simple Bayesian filtering techniques,” said Ahmad. “It can easily adapt to different road types and conditions, or different drivers using the same car.”

The research was conducted in collaboration with JLR who did the experimental design and the data collection. It was part of a project sponsored by JLR under the CAPE agreement with the University of Cambridge.

“This research is vital in understanding the impact of our design from a user perspective, so that we can continually improve safety and curate exceptional driving experiences for our clients,” said JLR’s Senior Technical Specialist of Human Machine Interface Dr Lee Skrypchuk. “These findings will help define how we use intelligent scheduling within our vehicles to ensure drivers receive the right notifications at the most appropriate time, allowing for seamless and effortless journeys.”

The research at Cambridge was carried out by a team of researchers from the Signal Processing and Communications Laboratory (SigProC), Department of Engineering, under the supervision of Professor Simon Godsill. It was led by Dr Bashar Ahmad and included Nermin Caber (PhD student at the time) and Dr Jiaming Liang, who all worked on the project while based at Cambridge’s Department of Engineering.

JOURNAL

IEEE Transactions on Intelligent Vehicles

DOI

10.1109/TIV.2023.3313419

ARTICLE TITLE

Driver Profiling and Bayesian Workload Estimation Using Naturalistic Peripheral Detection Study Data

New research demonstrates beef meals result in higher muscle protein synthesis rates than vegan meals

Evidence adds to growing body of research showing protein food sources are key to building and maintaining muscle

Peer-Reviewed Publication

MAASTRICHT UNIVERSITY MEDICAL CENTER

Long-standing research has shown that consuming dietary protein stimulates muscle protein synthesis, which is a critical factor for building and maintaining skeletal muscle mass. Growing evidence has illustrated that animal- and plant-based protein food sources are not created equal in terms of their anabolic properties for triggering muscle growth and maintenance, primarily due to the quantity and quality of protein in these foods, as well as their different essential amino acid (EAA) content.

New research recently published in the Journal of Nutrition is one of the first randomized controlled trials to compare anabolic properties of whole protein foods when consumed as part of mixed meals. The study, “Higher muscle protein synthesis rates following ingestion of an omnivorous meal compared with an isocaloric and isonitrogenous vegan meal in healthy, older adults,” found that, despite having the same caloric and total protein contents, a whole food omnivorous meal with lean beef resulted in greater postprandial muscle protein synthesis rates than a whole food vegan meal in older adults. In fact, researchers observed a 47% higher muscle protein synthesis rate following consumption of the omnivorous meal with lean beef, compared with the whole food vegan meal that provided an equal amount of protein from plants.

“While studies have previously assessed the impact of consuming isolated proteins, this research aims to mirror a more real-life setting by understanding the effects of eating whole protein foods as part of a typical meal,” said Luc van Loon, PhD, professor of Physiology of Exercise and Nutrition, Department of Human Biology, Maastricht University Medical Center+, and principal investigator of the research study. “Given the importance of protecting lean body mass to maintain strength as we age and the growing interest in vegetarian and vegan lifestyles, this research is important to understand if protein food sources can be equally effective in supporting muscle maintenance and growth.”

Based on previous research comparing the ingestion of different protein sources, the researchers were able to calculate that 16 participants would be needed to complete the study and detect a potential difference in muscle protein synthesis rates following ingestion of the two meals. Accordingly, the clinical trials were conducted with 16 healthy, older adults (ages 65-85 years), in Maastricht, the Netherlands. On one test day, the participants ate a whole food omnivorous meal containing 3.5 ounces of lean ground beef as the primary source of protein, with potatoes, string beans, applesauce (made of 100% apples), and herb butter. The other test day included eating a whole food vegan meal of equal caloric and protein content, comprised of unprocessed, commonly consumed plant protein foods such as quinoa, soybeans, chickpeas, and broad beans, as the main ingredients. Importantly, both meals contained on average 36 grams of protein, which is aligned with evidence-based recommendations for stimulating muscle protein synthesis in older individuals (i.e., 0.45 g protein per kg body weight).

“We were interested in studying the impact of meal consumption on muscle protein synthesis in older adults given the significance of age-related loss of muscle mass and strength, known as sarcopenia, which is a growing public health concern globally,” added van Loon.

All participants refrained from sports and strenuous physical activities, as well as alcohol consumption, for two days prior to each of the two experimental trial days. Researchers compared post-meal plasma amino acid profiles and muscle protein synthesis rates, using blood and muscle biopsies that were collected frequently for six hours following meal ingestion. In addition to observing the 47% increased muscle protein synthesis rate over a 6-hour postprandial period, researchers noted plasma EAA concentrations were 127% higher following the lean beef meal, despite the vegan meal not presenting any selective amino acid deficiencies.

“Importantly, plasma leucine, which is an essential amino acid particularly important for muscle protein synthesis, was 139% higher in participants, after they ate the omnivorous beef-containing meal,” said Philippe Pinckaers, MSc., lead author of the publication. “While more research is needed over a longer timeframe, this study illustrates the potential impact of the food matrix and significance of amino acid bioavailability and biofunctionality differences between beef-containing and vegan meals.”

The study was funded by the Beef Checkoff, through the National Cattlemen’s Beef Association, which was not involved in the study design, interpretation, or publication.

JOURNAL

Journal of Nutrition

DOI

10.1016/j.tjnut.2023.11.004

METHOD OF RESEARCH

Randomized controlled/clinical trial

SUBJECT OF RESEARCH

People

ARTICLE TITLE

Higher Muscle Protein Synthesis Rates Following Ingestion of an Omnivorous Meal Compared with an Isocaloric and Isonitrogenous Vegan Meal in Healthy, Older Adults

COI STATEMENT

The study was funded by the Beef Checkoff, through the National Cattlemen’s Beef Association, which was not involved in the study design, interpretation, or publication.

WIC participation helped families better cope with 2022 infant formula shortage

Peer-Reviewed Publication

WASHINGTON STATE UNIVERSITY

SPOKANE, Wash. – Families that participated in the WIC program—also known as the Special Supplemental Nutrition Program for Women, Infants and Children—were much less likely to use potentially unsafe infant feeding practices during the 2022 U.S. infant formula shortage than income-eligible families that did not participate.

Both WIC participants and non-participants reported being affected by the shortage at similar rates, according to a Washington State University study published in the Journal of the Academy of Nutrition and Dietetics. However, the researchers found that WIC participants were significantly more likely to cope with the shortage by changing the brand or type of formula or by getting it from a different source. They were also less likely to use less healthy feeding practices, such as using dairy milk or milk alternatives, watering down formula or using homemade formula.

“WIC provides a safety net for infants and children, and early childhood WIC participation has long-lasting benefits for health, wellbeing and academic achievement,” said Namrata Sanjeevi, the study’s first author and a research associate in WSU’s Elson S. Floyd College of Medicine. “By examining how WIC participation could be related to infant feeding practices during the formula shortage, our study adds important findings on how WIC can support families during times of crisis.”

The study evaluated how participation in WIC impacted families’ experiences and coping strategies during the monthslong shortage, which started in February 2022 when a manufacturer recall added to existing pandemic-related supply chain issues. A federal nutrition assistance program that provides free formula and essential nutritious foods to low-income families, the WIC program serves more than 1.4 million babies, about half of whom were receiving formula produced by the affected manufacturer.

Although data on why respondents selected specific coping strategies were not available, Sanjeevi said these findings provide some evidence that more flexible WIC policies may have eased the burden on participating families during the shortage. Following the manufacturer recall, the federal government temporarily waived restrictions on WIC benefits that limited participants as to the type, size and brand of formula they could obtain. Sanjeevi believes the improved access provided by these waivers—most of which were in place through at least the end of 2022—may have kept WIC participants from using undesirable feeding practices. In addition, she said the program could have protected participants from the budget impact of surging formula prices, since formula is provided to them free of cost.

Data for the study came from the Household Pulse Survey, an online survey designed to measure U.S. household experiences during the COVID-19 pandemic, and were collected between December 2022 and February 2023. The researchers’ analysis is based on data provided by 1,542 respondents whose household income was at or below 185% of federal poverty level—the threshold for WIC participation—and who had children younger than 18 months. Out of those respondents, 881, or just under 60%, reported participating in WIC.

Senior study author Pablo Monsivais, an associate professor in the WSU College of Medicine, said this is consistent with federal statistics that suggest that only about half of the women who are eligible for WIC are enrolled in the program.

“We need to do more to understand and eliminate barriers that keep families from participating in this proven, cost-effective program,” said Monsivais, adding that research has shown that every dollar invested in the program saves almost $2.50 in medical, educational and productivity costs.

“WIC has been put under the microscope again and again,” he said. “Our study adds to the growing body of evidence that the program protects the health and wellbeing of low-income families and makes good economic sense.”

JOURNAL

Journal of the Academy of Nutrition and Dietetics

DOI

10.1016/j.jand.2023.09.019

METHOD OF RESEARCH

Data/statistical analysis

SUBJECT OF RESEARCH

Not applicable

ARTICLE TITLE

WIC Participation and Coping Strategies Adopted by Families During the Infant Formula Shortage Crisis

“On-demand” HIV prevention method for women being tested in second early phase trial

Phase 1 study of the TAF/EVG fast-dissolving vaginal insert – intended for use at the time of sex – begins at US and African sites

Business Announcement

MATRIX: A USAID PROJECT TO ADVANCE THE RESEARCH AND DEVELOPMENT OF INNOVATIVE HIV PREVENTION PRODUCTS FOR WOMEN

PITTSBURGH – December 6, 2023 – A fast-dissolving vaginal insert that women would use at or around the time of sex as an “on-demand” HIV prevention method is being evaluated in a new early phase study being conducted by MATRIX, a United States Agency for International Development (USAID)-funded project focused on the early research and development of innovative HIV prevention products for women.

The insert, which resembles a bullet-shaped tablet, contains the antiretroviral (ARV) drugs tenofovir alafenamide (TAF) and elvitegravir (EVG). Once inside the vagina, it would begin to dissolve, and in doing so, release the two drugs. Animal and laboratory studies suggest the insert would provide protection against HIV for up to three days.

The MATRIX study is only the second Phase 1 trial of the TAF/EVG fast-dissolving insert used vaginally and the first to evaluate its use in multiple doses as well as in African women. The TAF/EVG fast-dissolving insert is the only on-demand HIV prevention product for use by women currently being evaluated in clinical trials.

The insert is being developed by CONRAD, a nonprofit research organization affiliated with Eastern Virginia Medical School (EVMS) in Norfolk, Va., U.S.A., for its use both vaginally and rectally. CONRAD-146, a first-in-human study conducted in the U.S. of its use as a vaginal insert, found that single administration was safe and acceptable among 16 women.

In the MATRIX study, known as MATRIX-001, researchers are evaluating the safety of the vaginal insert when used multiple times over several days, as well as user acceptability and how and where the two drugs are taken up in the body. The study, which will enroll 60 women at three sites in Kenya, South Africa and the United States, will help determine whether the product should advance to Phase 2 studies of its safety and acceptability when used as designed, i.e. at or around the time of sex.

Such a method could appeal to women who don’t want or are unable to use oral pre-exposure prophylaxis (PrEP), which requires taking an ARV tablet every day, or long-acting products like the monthly dapivirine vaginal ring or cabotegravir injections given every two months. It may be especially appealing to women who have infrequent or clustered sex and want only to use a product when needed, with local delivery (in the vagina) and with little drug going elsewhere in the body.

“Existing methods aren’t enough to meet women’s varying needs and lifestyles. A product that’s intended to be used at the time of sex, like the fast-dissolving TAF/EVG vaginal insert, would fill an important gap as alternative approach for women wanting protection only when they feel they need it,” said Nelly Mugo, MBChB, Mmed, MPH, MATRIX-001 protocol co-chair and investigator of record of the Kenya Medical Research Institute (KEMRI) Centre for Clinical Research Thika clinical research site (CRS), one of the three sites conducting the MATRIX-001study.

“Granted, the TAF/EVG insert is early in development, which is why the MATRIX-001 study is so critically important, especially for African women. This study will help determine the way forward for this product, and potentially get us one step closer to it being a viable option for women in this region,” added Leila Mansoor, PhD, protocol co-chair and the investigator of record at the Centre for the AIDS Programme of Research in South Africa (CAPRISA) eThekwini CRS, also a MATRIX-001 study site.

The TAV/EVG fast-dissolving insert contains 20 mg of TAF and 16 mg of EVG. TAF belongs to a class of ARVs called nucleoside reverse transcriptase inhibitors (NRTIs) that prevent HIV from making copies of itself inside human cells, therefore, preventing the spread of HIV inside the body. TAF has been approved by the U.S. Food and Drug Administration (FDA) for the treatment of chronic hepatitis B and for the treatment and prevention of HIV in men who have sex with men when used in combination with emtricitabine. Similarly, EVG has been approved by the U.S. FDA for the treatment of HIV in combination with other ARVs. EVG belongs to a class of ARV drugs known as integrase inhibitors that block HIV from being able to integrate its genetic code into human cells – a step that occurs later in the HIV lifecycle. Both TAF and EVG are being provided by Gilead Sciences for CONRAD’s development in the insert product.

Women in the study will be randomly assigned to use either the TAF/EVG fast-dissolving insert or a placebo insert with no active drugs. Each participant will use a total of 10 inserts – at first, every day for three consecutive days, and then every other day (every 48 hours) for two weeks. Participants will insert the products themselves, the first time in the clinic, with guidance from study staff. During the two to three months women are in the study, they will undergo different tests and procedures and will be asked questions about product acceptability prior to, during and following insert use. In addition, laboratory tests of tissue samples will be conducted to assess its potential activity against HIV, as well as herpes simplex virus (HSV), because pre-clinical laboratory and animal studies have shown that TAF acts against HSV in addition to HIV.

The study enrolled its first participants this week at the U.S. site, the EVMS CRS in Norfolk, Va, and the CAPRISA eThekwini CRS in South Africa has started screening potential participants. The study should be underway at the KEMRI Thika CRS in Kenya by early January. MATRIX-001 is expected to take approximately one year to conduct, with results anticipated mid-2025.

The TAF/EVG fast-dissolving insert is one of nine HIV prevention products being developed under MATRIX, and the only product to have previously been evaluated in clinical trials. In addition to the CONRAD-146 first-in-human study of its use vaginally, researchers have also conducted the MTN-039 first-in-human study of its use as a rectal insert, which found single use and two inserts used together posing no safety concerns. In both studies, results of laboratory tests of tissue and fluid samples showed drug levels compatible with protection against HIV.

According to UNAIDS, women and girls accounted for 63 percent of all new HIV infections in sub-Saharan Africa in 2022, versus 46 percent globally. In much of Africa, daily oral PrEP is the only biomedical prevention method available, and daily pill-taking has been especially challenging for adolescent girls and young women. Both the monthly dapivirine ring and cabotegravir long-acting injectable (CAB-LA) have been recommended by the World Health Organization and approved for use in several African countries, though neither method is yet to be made widely available. Even so, women have different preferences and needs, and at different times in their lives, which is why additional options are needed.

MATRIX is a five-year program funded by USAID in 2021 that aims to expedite the research and development of HIV prevention products for women – including products designed to protect against both HIV and pregnancy – that in addition to being safe and effective, will be acceptable, affordable, scalable and deliverable in the settings where they are needed most. MATRIX activities are focused on the early research and development of products, which involves both pre-clinical research and the first clinical trials of products. Through its North-South partnerships, MATRIX also aims to strengthen the capacity of African investigators to facilitate full and sustainable ownership of this work into the future. MATRIX is being implemented by Magee-Womens Research Institute (MWRI) in collaboration with partner organizations based in Kenya, South Africa, the United States and Zimbabwe. Leading the project is Sharon Hillier, Ph.D., of MWRI and the University of Pittsburgh School of Medicine, with Thesla Palanee-Phillips, Ph.D., from the Wits RHI and University of Witwatersrand, South Africa, serving as deputy director.

# # #

To learn about MATRIX go to www.matrix4prevention.org. Click here to read a QA about the TAF/EVG fast-dissolving vaginal insert and MATRIX-001 study. Additional information about MATRIX-001 can also be found at https://www.matrix4prevention.org/activity-hubs/clinical-trials/matrix-001.

MATRIX was established through the generous support of the American people through the U.S. President’s Emergency Plan for AIDS Relief (PEPFAR) and the U.S. Agency for International Development (USAID).

(United States Agency for International Development (USAID) Cooperative Agreement Number 7200AA22CA00002)

The content and views in this document are those of MATRIX and its partners and do not necessarily reflect the views of PEPFAR, USAID or the U.S. Government.

METHOD OF RESEARCH

Randomized controlled/clinical trial

SUBJECT OF RESEARCH

People

Disclaimer:

Having a C-section is related to difficulties with conceiving

Peer-Reviewed Publication

THE UNIVERSITY OF BERGEN

Previous studies shown that women who have had a C-section tend to have more problems conceiving a baby than ones who have had normal, vaginal birth.

“Many of these studies have utilized inter-pregnancy intervals to measure women’s fertility,” researcher Yeneabeba Sima at the University of Bergen, explains. She points out:

“However, a measure of inter-pregnancy interval cannot distinguish between voluntary and involuntary delay in getting pregnant.”

Asking women if they planned their pregnancies.

Using data from The Norwegian Mother, Father and Child Cohort Study (MoBa) linked to the Medical Birth Registry of Norway (MBRN), Sima and colleagues assessed women’s fertility.

The MoBa questionnaire inquired whether or not women planned their pregnancies.

“For those who actively tried to have a baby, we examined the time it took for them to conceive. If they had tried for a year or more before getting pregnant, they were considered to have reduced fertility,” says Sima.

The researchers examined differences in time spent trying to conceive among 42,379 MoBa participants, all of whom had at least one previously registered birth in the MBRN. The findings indicated that women with a prior C-section had a 10% decreased chance of conceiving their next pregnancy during a given menstrual cycle compared with those who had prior vaginal deliveries.

Women with fertility problems also had more C-sections.

The researchers also explored the association in the other direction, between reduced fertility and later C-section. Among 74,025 MoBa participants, 11% reported trying for more than one year before getting pregnant. They found that women who took one year or longer to conceive were 21% more likely to be delivered by C-section, as compared with women who spent less than 12 months trying to conceive.

“In our study, women with difficulty conceiving have a higher prevalence of pregnancy complications. There is also a higher prevalence of chronic health issues like diabetes mellitus and high blood pressure among these women. However, the increased risk of having a C-section still existed for women who didn't have these health issues," said Sima.

The associations between C-section and reduced fertility might not be causal.

Previous studies concluded that reduced fertility following C-section could be a side effect of the surgical operation. However, Sima and colleagues suggest that common underlying risk factors could contribute to both reduced fertility and C-section:

“Maternal stress might be one reasonable explanation connecting challenges in conceiving and an elevated risk of labor difficulties, ultimately leading to a higher likelihood of C-section,” Sima explains, and adds:

“Our findings suggest that the observed reduced ability to conceive after C-section may be linked to underlying maternal conditions not registered in our data or not yet clinically emerged, and the surgical procedure may not directly influence this pathway.”

JOURNAL

American Journal of Obstetrics and Gynecology

DOI

10.1016/j.ajog.2023.10.029

METHOD OF RESEARCH

Survey

SUBJECT OF RESEARCH

People

ARTICLE TITLE

The relationship between cesarean delivery and fecundability: a population-based cohort study

How ChatGPT could help first responders during natural disasters

UB researchers train AI to accurately recognize addresses and other location descriptions in Hurricane Harvey social media posts

Peer-Reviewed Publication

UNIVERSITY AT BUFFALO

BUFFALO, N.Y. — A little over a year since its launch, ChatGPT’s abilities are well known. The machine learning model can write a decent college-level essay and hold a conversation in an almost human-like way.

But could its language skills also help first responders find those in distress during a natural disaster?

A new University at Buffalo-led study trains ChatGPT to recognize locations, from home addresses to intersections, in disaster victims’ social media posts.

Supplied with carefully constructed prompts, researchers’ “geoknowledge-guided” GPT models extracted location data from tweets sent during Hurricane Harvey at an accuracy rate 76% better than default GPT models.

“This use of AI technology may be able to help first responders reach victims more quickly and even save more lives,” said Yingjie Hu, associate professor in the UB Department of Geography, within the College of Arts and Sciences, and lead author of the study, which was published in October in the International Journal of Geographical Information Science.

Disaster victims have frequently turned to social media to plead for help when 911 systems become overloaded, including during Harvey’s devastation of the Houston area in 2017.

Yet first responders often don’t have the resources to monitor social media feeds during a disaster, following the various hashtags and deciding which posts are most urgent.

It is the hope of the UB-led research team, which also includes collaborators from the University of Georgia, Stanford University and Google, that their work could lead to AI systems that automatically process social media data for emergency services.

“ChatGPT and other large language models have drawn controversy for their potential negative uses, whether it be academic fraud or eliminating jobs, so it is exciting to instead harness their powers for social good,” Hu says.

"While there are a number of significant and valid concerns about the emergence of ChatGPT, our work shows that careful, interdisciplinary work can produce applications of this technology that can provide tangible benefits to society,” adds co-author Kenneth Joseph, assistant professor in the UB Department of Computer Science and Engineering, within the School of Engineering and Applied Sciences.

Fusing ‘geoknowledge’ into ChatGPT

Imagine a tweet with an urgent but clear message: A family, including a 90-year-old not steady on their feet, needs rescuing at 1280 Grant St., Cypress, Texas, 77249.

A typical model, such as a named entity recognition (NER) tool, would recognize the listed address as three separate entities — Grant Street, Cypress and Texas. If this data was used to geolocate, the model would send first responders not to 1280 Grant St., but into the middle of Grant Street, or even the geographical center of Texas.

Hu says that NER tools can be trained to recognize complete location descriptions, but it would require a large dataset of accurately labeled location descriptions specific to a given local area, a labor-intensive and time-consuming process.

“Although there’s a lack of labeled datasets, first responders have a lot of knowledge about the way locations are described in their local area, whether it be the name of a restaurant or a popular intersection,” Hu says. “So we asked ourselves: How can we quickly and efficiently infuse this geoknowledge into a machine learning model?”

The answer was OpenAI’s Generative Pretrained Transformers, or GPT, large language models already trained from billions of webpages and able to generate human-like responses. Through simple conversation and the right prompts, Hu’s team thought GPT could quickly learn to accurately interpret location data from social media posts.

First, researchers provided GPT with 22 real tweets from Hurricane Harvey victims, which they’d already collected and labeled in a previous study. They told GPT which words in the post described a location and what kind of location it was describing, whether it be an address, street, intersection, business or landmark.

Researchers then tested the geoknowledge-guided GPT on another 978 Hurricane Harvey tweets, and asked it to extract the location words and guess the location category by itself.

The results: The geoknowledge-guided GPT models were 76% better at recognizing location descriptions than GPT models not provided with geoknowledge, as well as 40% better than NER tools. The best performers were the geoknowledge-guided GPT-3 and GPT-4, with the geoknowledge-guided ChatGPT only slightly behind.

“GPT basically combines the vast amount of text it’s already read with the specific geoknowledge examples we provided to form its answers,” Hu says. “GPT has the ability to quickly learn and quickly adapt to a problem.”

However, the human touch, that is, providing a good prompt, is crucial. For example, GPT may not consider a stretch of highway between two specific exits as a location unless specifically prompted to do so.

“This emphasizes the importance of us as researchers instructing GPT as accurately and comprehensively as possible so it can deliver the results that we require,” Hu says.

Letting first responders do what they do best

Hu’s team began their work in early 2022 with GPT-2 and GPT-3, and later included GPT-4 and ChatGPT after those models launched in late 2022 and early 2023, respectively.

“Our method will likely be applicable to the newer GPT models that may come out in the following years,” Hu says.

Further research will have to be done to use GPT’s extracted location descriptions to actually geolocate victims, and perhaps figure out ways to filter out irrelevant or false posts about a disaster.

Hu hopes their efforts can simplify the use of AI technologies so that emergency managers don’t have to become AI experts themselves in order to use these them, and can focus on saving lives.

“I think a good way for humans to collaborate with AI is to let each of us focus on what we're really good at,” Hu says. “Let AI models help us complete those more labor-intensive tasks, while we humans focus on gaining knowledge and using such knowledge to guide AI models.”

The work was supported by the National Science Foundation.

JOURNAL

International Journal of Geographical Information Science

DOI

10.1080/13658816.2023.2266495

METHOD OF RESEARCH

Computational simulation/modeling

ARTICLE TITLE

Geo-knowledge-guided GPT models improve the extraction of location descriptions from disaster-related social media messages

Ex-entrepreneurs can thrive in the right employee roles, UCF researcher finds in new study

Assistant Professor Jeff Gish co-authored a study which found that former entrepreneurs can successfully transition into employees, especially in roles that harness their entrepreneurial spirit.

Peer-Reviewed Publication

UNIVERSITY OF CENTRAL FLORIDA

Ex-entrepreneurs Can Thrive in the Right Employee Roles — IMAGE:
IN THE STUDY, RESEARCHERS EXAMINED THE IDENTITY CONFLICT LEVELS OF FORMER ENTREPRENEURS WHO WENT ON TO WORK FOR AN ORGANIZATION. IMAGE CREDIT: ANTOINE HART
view more
CREDIT: ANTOINE HART/UNIVERSITY OF CENTRAL FLORIDA

ORLANDO, Dec. 7, 2023 — Once an entrepreneur always an entrepreneur? Not necessarily, says a new study by researchers at the University of Central Florida and Purdue University. Former entrepreneurs can transition from being their own boss into successful employees within an organization, especially in roles that harness their entrepreneurial spirit, according to a recent study published in Personnel Psychology.

“With today’s career paths typically spanning multiple roles across a variety of organizations, understanding the transition between someone’s old work self and new work self may be critical to not only the employee’s success but also the company’s,” says Jeff Gish, assistant professor of management and entrepreneurship in UCF’s College of Business and the study’s co-author.

Gish and co-author Jordan Nielsen, an assistant professor of management organizational behavior/human resources at Purdue, examined the identity conflict levels of former entrepreneurs who went on to work for an organization.

Research has shown that former entrepreneurs frequently experience a “founder penalty” when applying for jobs, losing out to applicants who have never been self-employed. Employers assume former entrepreneurs may be more difficult to manage or will jump ship to start another company and be their own boss again. This new research suggests that this need not be the case for all jobs or for all ex-entrepreneurs.

They surveyed ex-entrepreneurs about their current work identity and whether they felt they could act like an entrepreneur in their current work role or if they had to suppress their entrepreneurial spirit. They also surveyed the ex-entrepreneurs’ romantic partners about whether the employee spoke highly of their current organization, engaging in boosterism or experienced burnout in the role.

Gish and Nielsen found that identity conflict between the old entrepreneurial self and the new employee self was associated with higher levels of burnout and lower levels of boosterism.

“Ex-entrepreneurs who felt a strong nostalgia for being their own boss tended to be the ones who were the most negatively affected, with the highest levels of burnout and lowest levels of boosterism,” Nielsen says. “To mitigate this, organizations could use interview questions to help identify those who may be more likely to suffer negative consequences or develop positions and onboarding practices that minimize this source of conflict and lay a stronger foundation for success.”

JOURNAL

Personnel Psychology

DOI

10.1111/peps.12626

METHOD OF RESEARCH

Survey

SUBJECT OF RESEARCH

People

ARTICLE TITLE

When old and new selves collide: Identity conflict and entrepreneurial nostalgia among ex-entrepreneurs

Petty Bourgeoisie and Middle Class. The lower middle class or the petty (petite) bourgeoisie (the bourgeoisie was sometimes called the middle class in this era), “are smallowners who still work their own means of production, or owner-workers” (Adams and Sydie, p. 134). The characteristic of this class is that it does own some property, but not sufficient to have all work done by employees or workers. Members of this class must also work in order to survive, so they have a dual existence – as (small scale) property owners and as workers. Because of this dual role, members of this class have divided interests, usually wishing to preserve private property and property rights, but with interests often opposed to those of the capitalist class. This class is split internally as well, being geographically, industrially, and politically dispersed, so that it is difficult for it to act as a class. Marx expected that this class would disappear as capitalism developed, with members moving into the bourgeoisie or into the working class, depending on whether or not they were successful. Many in this class have done this, but at the same time, this class seems to keep recreating itself in different forms.

Marx considers the petite bourgeoisie to be politically conservative or reactionary, preferring to return to an older order. This class has been considered by some Marxists to have been the base of fascism in the 1920s and 1930s. At other times, when the petty bourgeoisie is acting in opposition to the interests of large capital, it may have a more radical or reformist bent to it – for example, small employers may express opposition to large monopolies that may be attempting to eliminate small businesses.

Note on the Middle Class. The issue of the middle class or classes appears to be a major issue within Marxian theory, one often addressed by later Marxists. Many Marxists attempt to show that the middle class is declining, and polarization of society into two classes is a strong tendency within capitalism. Marx's view was that the successful members of the middle class would become members of the bourgeoisie, while the unsuccessful would be forced into the proletariat. In the last few years, many have argued that in North America, and perhaps on a world scale, there is an increasing gap between rich and poor and there is a declining middle.

While there have been tendencies in this direction, especially among the farmers and peasantry, there has been no clear long run trend toward decline of the middle class. At the same time as there has been polarization of classes, there have been new middle groupings created. Some of these are small business people, shopkeepers, and small producers while others are professional and managerial personnel, and some intellectual personnel. Well paid working class members and independent trades people might consider themselves to be members of the middle class. Some segments of this grouping have expanded in number in recent years. While it is not clear that these groups hold together and constitute a class in any Marxian sense of being combined in opposition to other classes, they do form a middle grouping. Since Marx's prediction has not come true, sociologists and other writers have devoted much attention to explaining this middle grouping – what is its basis, what are the causes of its stability or growth, how it fits into the class structure, and what are the effects of its existence on proletariat and bourgeoisie.

Marx on Social Class

University of Regina

https://uregina.ca › ~gingrich

Jan 31, 2003 — Marx considers the petite bourgeoisie to be politically conservative or reactionary, preferring to return to an older order. This class has ...