Tag: PLOS

Jolting Awake Often is Associated with Poorer Sleep and Higher Anxiety, Survey Finds

An online survey of 194 adults in Austria and Germany compared those who startle awake at least weekly with those who never or only rarely do.

Photo by Andrea Piacquadio

People who frequently startle awake while falling asleep – a phenomenon researchers call ‘sleep starts’ or hypnic jerks – report poorer sleep quality, greater insomnia severity, and higher anxiety than those who do not, according to a new study published this week in the open-access journal PLOS Mental Health by Tanja Grünberger and Manuel Schabus of Paris Lodron University Salzburg, Austria, and colleagues.

Sleep starts are sudden jolts that occur in the moments between wakefulness and sleep, most often felt as an abrupt muscle jerk or twitch, sometimes strong enough to jolt a person fully awake. The experience can also involve a sudden sensation of falling or loss of balance, electric-shock-like feelings, tingling or numbness, flashes of light, or brief, dream-like imagery. These experiences are frequently accompanied by a rush of emotion – alarm, confusion, fear, or panic – and physical signs of a startle response, such as a racing heart or rapid breathing. Afterward, most people fall back asleep quickly.

In the new study, researchers conducted an online survey of 194 adults in Austria and Germany who answered questions about their sleep quality, insomnia severity, daytime sleepiness, dysfunctional sleep-related beliefs, and symptoms of anxiety and depression. 96 of the participants experienced abrupt awakenings and at least one other symptom weekly, while the other 98 were used as a control group.

Compared with the control group, the sleep-starts group reported poorer sleep quality, greater insomnia severity, more dysfunctional sleep-related beliefs, and higher anxiety and depressive symptoms than those in the control group (all statistically significant). How often people experienced sleep starts was also linked to sleep quality, sleep beliefs, and anxiety, but not depression. Together, these factors, along with age and gender, were associated with about a quarter of the differences between people in sleep start frequency (26.9%). Participants also reported more frequent sleep starts during periods of psychological stress and emotional strain, whereas sleep deprivation was less influential on the frequency of sleep starts.

“These findings support the view that sleep starts should be considered within broader frameworks of sleep-related hyperarousal and cognitive-emotional activation,” the authors write. “Although causality cannot be inferred from these cross-sectional self-report data, the results highlight potentially relevant links between sleep starts and sleep-related cognitive-emotional processes.”

“These findings support the view that sleep starts should be considered within broader frameworks of sleep-related hyperarousal and cognitive-emotional activation,” the authors write. “Although causality cannot be inferred from these cross-sectional self-report data, the results highlight potentially relevant links between sleep starts and sleep-related cognitive-emotional processes.” First author Tanja Grünberger says: “Because there is no established questionnaire for sleep starts, we developed our own items based on the clinical literature. This allowed us to capture a broad range of experiences – but it also means that our findings now need to be confirmed with validated instruments and objective sleep recordings.” 

Dr Grünberger adds: “Sleep starts are much more than a simple muscle twitch when falling asleep. Participants described not only jerks, but also sensations of falling, a racing heart, flashes of light or loud bangs, and strong feelings of alarm or fear – and those who reported sleep starts at least once a week also reported poorer sleep quality, more unhelpful beliefs about sleep and more anxiety symptoms.”

She continues: “What we found most interesting was how closely sleep starts are linked to how people think and feel about sleep. The more strongly participants held unhelpful beliefs about sleep and the more anxious they were, the more often they reported sleep starts – and in the moment itself, many described a strong pressure to finally fall asleep. Our data cannot show what causes what, but this pattern resembles what we know from insomnia.”

 Senior author Manuel Schabus explains prospective next steps for their research: “For most people, sleep starts are a harmless feature of falling asleep. For a subgroup, however, they go along with distress and a longer time to fall asleep. Whether approaches that reduce arousal and sleep-related worry also help with distressing sleep starts has not yet been tested – that is an important next step.”

Provided by PLOS

Many Non-drug and Non-surgical Therapies for Low Back Pain are Cost-effective

Interventions like yoga, cognitive therapy, and self-management support were more cost-effective than standard medical advice

Photo by RDNE Stock project

Non-drug and non-surgical treatments for low back pain, including psychological support, exercise and self-management strategies are cost-effective interventions, according to a study published October 27th in the open access journal PLOS Medicine by Manuela Loureiro Ferreira from the George Institute for Global Health, Australia, and colleagues.

Low back pain is the leading cause of disability worldwide, with significant costs to both individuals and healthcare systems. There are pharmacological or surgical interventions – including drug prescriptions, spinal injections and surgery, but non-invasive methods such as cognitive therapy, physical therapy, exercise and yoga or dietary interventions are also available. Understanding which options provide best value for the money is important for determining where to allocate healthcare funding.

To better understand the cost-effectiveness of non-invasive interventions for low back pain, the authors of this study reviewed research studying treatments recommended by the World Health Organization, the United Kingdom’s National Institute for Health and Care Excellence (NICE), and the North American Spine Society. They included 20 studies conducted between 2005 and 2026, with a total of 14,884 participants. Those studies compared non-pharmacological and non-surgical interventions such as yoga, acupuncture, massage, psychological therapy, spinal manipulation, exercise and diet plans to some first contact care or no treatment.  

The authors found that six single therapies – yoga, cognitive behavioural therapy, self-management support, fear reduction education and mindfulness stress reduction, could reduce health care costs and improve outcomes compared with usual care. They also observed that two multi-modal therapies – combining advice and graded activity in the workplace, and diet plus physical activity – were more cost-effective than usual care, at a willingness to pay threshold of £20 000 (R440 000; meaning that costs were below that maximum price, a threshold recommended by the NICE). While all of the studies examined were from high-income countries and there were differences in how cost effectiveness was measured, the authors suggest that improving access to these non-surgical and non-pharmacological treatments could benefit patients with low back pain.

Senior author Prof Manuela Ferreira adds, “Low back pain is the world’s leading cause of disability, but many effective treatments can also save healthcare dollars. Our review found that approaches such as exercise, yoga, acupuncture and psychological support can help people recover while providing good value for money.

“The good news for people with back pain is that some of the most effective treatments are also among the most cost-effective. Investing in evidence-based care such as exercise, physical therapies and psychological support could help more people get better while reducing pressure on health systems.”

Provided by PLOS

Opioids During Pregnancy Linked to Higher Rates of Childhood Hospital Admission

A study of nearly 3 million South Korean children found a 28% higher rate of hospital admission among those exposed prenatally to opioids

Photo by SHVETS production

Children whose mothers received opioid prescriptions during pregnancy have a higher rate of hospital admission in childhood than unexposed children, including unexposed siblings, according to a new study published October 13th in the open access journal PLOS Medicine by Dong Keon Yon of Kyung Hee University College of Medicine, South Korea, and colleagues.

Over recent decades, opioid prescriptions and doses have risen in clinical practice, and a considerable proportion of pregnant women are exposed to these drugs. In the US, for example, data has shown that 14-22% of women filled a prescription for opioids during pregnancy. Prenatal opioid exposure may disrupt infant development and immune system function, but prior studies have had limitations such as confounding factors, limited sample sizes, or restricted follow-up durations.

In the new study, researchers used nationwide health insurance data on mothers and children in South Korea, following a cohort of 2,928,186 children born between 2010 and 2017 for up to 14 years. Children were classified as prenatally exposed to opioids if their mothers received two or more opioid prescriptions during pregnancy, and were compared both with unexposed children and with their own siblings.

Children with prenatal opioid exposure had a 28% higher rate of hospital admission than unexposed children (adjusted incidence rate ratio 1.28, 95% CI 1.28-1.29). The association was smaller but still significant when comparing siblings (aIRR 1.08, 95% CI 1.07-1.10). The effect was also dose-dependent, with higher maternal opioid doses associated with higher admission rates (aIRR 1.19 for low-dose exposure versus 1.40 for high-dose exposure). The elevated admission rates spanned several causes, including infections and respiratory, gastrointestinal and genitourinary conditions, with the association strongest up to five years of age.

“These findings suggest the need for additional postnatal health checks for those born to mothers with opioid prescriptions during pregnancy,” the authors write. “However, these findings should be interpreted with caution, as the study relied on electronic health records, which may not reflect actual use of medication, and unmeasured factors shared within families may still partly explain the observed associations.”

Dong Keon Yon, MD, PhD (co-corresponding author) states, “In a nationwide cohort of nearly three million Korean children, those whose mothers received two or more opioid prescriptions during pregnancy had about a 28% higher rate of hospital admission, most notably for infections and respiratory illness in the first years of life. The risk rose with higher opioid doses and longer prescriptions, which points to a possible biological effect—but when we compared siblings within the same family, the association shrank considerably, so shared family factors clearly explain part of what we see.”

Jiseung Kang, PhD (co-corresponding author), adds, “Because opioids in South Korea are obtained almost entirely by prescription, our data let us examine prescribed opioid exposure in pregnancy unusually cleanly, and following nearly three million children for up to 14 years gave us the statistical power to look at dose, timing, and specific causes of admission separately. The consistency of the pattern across all of these analyses is what makes the finding hard to dismiss.” 

Hyeon Jin Kim, MS (joint first author) states, “Our message is not that pregnant women should never receive opioids—for some, pain relief is essential. It is that opioids in pregnancy should be prescribed at the lowest effective dose for the shortest time, and that children born after such exposure may benefit from closer routine health checks, particularly in infancy when the risk was highest.”

Yesol Yim (joint first author) notes, “What surprised us most was how quickly the association faded as children grew older. The excess risk was clearest in the first year of life and had largely disappeared after age five. That pattern suggests a window of vulnerability in early infancy rather than a lifelong effect – and it is a window in which pediatric follow-up can make a real difference.”

Provided by PLOS

Variations in the Human Vocal Tract Could Have Influenced Our Language Diversity

Vocal tracts and the sounds they produce vary, resulting in distinct sounds in speaking groups

Vocal tract outlines for the five vowels [i], [e], [a], [o] and [u] in the midsagittal plane producing a vowel sound. Each colour represents a different vowel and speaker, showcasing markedly different vocal tract and tongue configurations across speakers and vowels.
Image credit: Tiena Danner and Christoph Zollikofer (CC-BY 4.0)

The anatomical differences in our vocal tracts could be one of the sources of our Tower of Babel – the vast difference in languages spoken around the world, according to a study published October 13th in the open access journal PLOS Biology by Steven Moran and Tiena Danner from the University of Neuchâtel, Switzerland, and colleagues.

Many variations exist in the human voice. People vary in the size and shape of the tissues in the head and neck and also shape that anatomy differently to produce similar sounds. Developmental and cultural factors will shape the way people speak. To better understand how anatomical and articulatory variation might affect how a speaker sounds, the authors of this study analysed data from 32 speakers (16 male and 16 female) from the USC 75-Speaker Speech MRI Database, an open-source collection of real-time Magnetic Resonance Imaging (rtMRI) conducted during speech tasks, focusing on American English vowels (i, e, a, o and u).

The authors found that the subjects had widely different vocal tracts and used very different ranges of motion in those tracts to produce the same sound. Those ranges of motion, or articulatory differences, were a result of each person’s specific vocal anatomy, and produced both the correct sound and an individual acoustic difference. Finally, by combining the rtMRI results with multilingual speech data, they showed that the amount of variation within other languages such as Basque or Hindi was similar to the variation present in American English, suggesting that within each language, there’s a basic level of speaker variation as a result of vocal tract anatomy.

While the sample size of vocal tracts was small, and the authors compared variations in only a few language families, the authors propose a biocultural model of how spoken sounds might vary and diversity between peoples, where variation of pronunciation within a group feeds into the development of a language difference between groups, eventually producing two groups that can’t understand each other.

The authors add, “We often think of differences in pronunciation between speakers of a language as noise to be normalized away in linguistic analysis. However, the results of our study suggest the opposite: that the variation between individuals may be an important source of information in understanding the evolution of spoken languages.

“An exciting aspect of our study is that it provides researchers with a null model for studying the phonetic evolution of spoken languages. If the differences between the world’s 7000 or so languages reflect variation already present in the anatomical differences between speakers, we now have a baseline against which we can identify sociocultural or historical factors that have driven language diversification.

“Every person has a slightly different vocal tract and speakers compensate for these differences to remain mutually intelligible. Nevertheless, our anatomical differences leave systematic acoustic signatures in the sounds we produce. Remarkably, we find that the same patterns of phonetic variation between individuals are also reflected in the difference between languages.

“Our results suggest that the seeds of phonetic differences between languages are already present in the variation between individual speakers. Spoken language evolution is driven by the biological diversity within populations and the sociocultural forces that shape how languages change.”

Provided by PLOS

Total Ultra-processed Food Intake not Linked to Colorectal Cancer in Large Study

Total ultra-processed food intake not linked to colorectal cancer in large US cohort study

Overall ultra-processed food (UPF) consumption was not associated with increased colorectal cancer (CRC) incidence in a cohort of more than 400 000 US adults, according to a study published on 15 October by Erikka Loftfield of the National Cancer Institute, US, and colleagues in the open access journal PLOS Medicine.

Ultra-processed foods account for more than half of energy intake among US adults and have attracted growing attention due to potential links to chronic diseases, including colorectal cancer. However, UPFs represent a wide range of foods. Some, like whole grain and dairy products, are associated with a lower risk of CRC while others, like processed meats, are associated with a higher risk.

To better understand these relationships, researchers analysed data from 461 682 adults aged 50 to 71 years enrolled in the NIH-AARP Diet and Health Study between 1995 and 1996. Participants recorded their eating habits using a food frequency questionnaire that researchers used to estimate UPF intake. Participants were then followed until 2018, a median of 18.8 years.

The study found no association between overall UPF intake and CRC incidence. However, findings differed by type of food. Higher intake of whole grains, fibre, dairy products, and calcium derived from UPF sources was associated with lower CRC incidence, whereas meat from UPF sources was associated with higher incidence. Researchers also found that replacing 10% of gram weight from UPF intake with minimally processed foods was associated with a 2% lower CRC incidence.

The researchers caution against generalising these findings to younger, more diverse populations. The study cohort included primarily older, white Americans and only measured UPF intake once at baseline. As a result, the findings cannot account for changing eating habits or the increasing availability in UPFs, potentially leading to an underestimation of actual UPF intake.

The study suggests that UPFs should not be treated as a monolith and reinforce current dietary guidance emphasising whole grains, fibre, dairy, and calcium while limiting processed meat intake to reduce the risk of CRC.

Loftfield adds, “In this large population of older US adults, we did not observe an association between total ultra-processed food (UPF) intake and colorectal cancer incidence; however, when we looked at food groups, like whole grains, dairy, and meat, by UPF or non-UPF source, we found that associations were in line with current dietary guidance for colorectal cancer prevention such that higher whole grain and dairy but lower meat intakes were associated with lower colorectal cancer incidence.”

Provided by PLOS

Skin Sensors Show Link Between Insomnia and Fear Responses

A small study measuring stress-related skin signals found differences in the learned fear and safety responses of adults with and without insomnia.  

Demonstration of skin conductance sensors attached to the palm, used for estimating physiological arousal. Image credit: Revanth Reddy, CC-BY 4.0

A new computational model using skin conductance measurements to estimate stress-related emotional arousal found a significant difference in learned fear and safety responses over time between adults with and without insomnia. The new study from Dr. Rose T. Faghih of New York University Tandon School of Engineering, US, and collaborators at Harvard Medical School, US, describing the work was published this week in the open-access journal PLOS Mental Health.

Insomnia Disorder, affecting an estimated 10% of the population, has been widely conceptualised as a disorder of chronic hyperarousal linked to emotional and autonomic dysregulation. Sleep is thought to help consolidate fear extinction, the process of learning that a previously threatening stimulus no longer signals danger, forming a new, inhibitory “safety memory” to override an established “fear memory.” 

 Skin conductance, changes in the skin’s electrical properties driven by sweat gland activity, is a well-established marker of stress and emotional arousal. Previous studies looking at skin conductance and fear extinction have focused primarily on sharp, isolated incidences of stress activity.

 In the new study, researchers designed two statistical models estimating stress and arousal from skin conductance: one model using only isolated stress spikes, and one model that also incorporated the signal’s continuous background levels. Both models were applied to skin conductance recordings from an initial cohort of 24 adults with insomnia and 24 good sleepers who completed a task in which they were temporarily conditioned to fear a visual stimulus, then learned it was no longer dangerous.

 Only the model including the tonic background signal detected a significant difference between the two groups. From the conditioning phase, when the visual stimulus was paired with a mild shock, to the extinction phase, when it no longer was, the average ‘high-arousal index’ increased in people with insomnia but slightly decreased in good sleepers. Other differences between the groups — including lower average arousal during conditioning and higher average arousal when the fear and safety memories were recalled on the second day — followed a similar pattern but did not reach statistical significance.

 These findings suggest that because people with insomnia failed to fully acquire the safety memory on the first day or retain it until the second day, the new extinction memory was too weak to oppose the original fear memory during recall. Lead author Tianyu Hwang summarises the experiment as follows: “Our study shows that insomnia isn’t just about poor sleep – it directly relates to the body’s ability to learn and remember that a threat is gone.” 

 “Hyperarousal is thought to be an underlying risk factor for developing insomnia, and we were excited to see that sophisticated mathematical parsing of skin conductance revealed the same effect we had previously observed using fMRI,” co-author Dr Edward Pace-Schott writes. “This hyperarousal could oppose subsequent learning that a stimulus has become safe, which may explain why anxiety treatments like exposure therapy are more successful in people with good sleep.”

 Senior author Dr Rose Faghih adds: “Using our previously developed sympathetic arousal virtual sensors based on microfluctuations in sweat activity, we compared insomnia patients and healthy controls during fear extinction and observed that insomnia is associated with a reduced capacity to extinguish physiological threat responses.” 

 Dr Ryan Bottary, co-author, notes: “In our study, people with insomnia, compared to their good sleeping peers, became more aroused once the danger was taken away. For patients, this may help explain why the fear that ‘a bad night will ruin tomorrow’ survives so many tomorrows that turn out just fine.”

 Co-author Dr Saman Khazaei concludes: “I find it interesting to see that decoded arousal from certain components of skin conductance reflects a signature of a disrupted fear response in people diagnosed with insomnia. This finding may motivate further research on skin conductance and the use of wearable devices that measure it to improve sleep in everyday-life settings.”

Provided by PLOS

Blood Test Trends may Help Identify Patients at Increased Risk of Cancer

Among patients with unexplained weight loss, changes over time in routine blood test results were associated with overall and site-specific cancer diagnoses

Photo: Pixabay

Blood test trends alongside unexplained weight loss can improve triage for cancer testing, according to a study by Brian Nicholson and colleagues from the University of Oxford, UK, published September 17th in the open access journal PLOS Medicine.

Abnormal blood test results, such as low haemoglobin or increased platelet counts, can provide clues about a patient’s risk of developing cancer. However, monitoring trends over repeat tests could provide further clues in some instances, by identifying cancer-related changes in blood test results that do not appear abnormal. Assessing patterns in blood test abnormalities and trends alongside unexplained weight loss could enhance cancer risk assessment and support earlier diagnosis.

In this study, researchers examined trends in blood test results among patients with unexplained weight loss, a common non-specific symptom associated with multiple cancer types. The goal was to determine whether trends in 26 commonly used blood tests in primary care could improve cancer risk stratification compared to abnormalities on single blood tests.

Among the more than 275 000 patients included in the study, nearly 14 000 were subsequently diagnosed with cancer within six months, allowing researchers to relate abnormalities on single tests and trends over repeat blood tests to cancer diagnosis. Overall, 23 blood test trends were associated with overall cancer risk. Several abnormalities were also associated with specific cancer types. After accounting for age and sex, several blood test trends were more discriminative for cancer diagnosis than individual abnormal test results. For example, trends in white blood count and neutrophils were linked to lung cancer while trends in red blood cell count, haematocrit, and platelet-to-lymphocyte ratio were associated with prostate cancer.

The findings suggest that monitoring changes in routine blood test results in patients with unexplained weight loss could identify patients in primary care who would benefit from additional cancer testing.

Author Brian Nicholson adds, “We show how blood test results are made more accurate for cancer by adding patient age and sex. This relatively simple calculation could easily be performed at the laboratory.”

Author Pradeep Virdee states, “In some instances, monitoring how a patient’s blood test results change over time could offer further value. We plan to assess how well blood test abnormalities and trends inform cancer risk in patients with other types of non-specific symptoms, such as fatigue and vomiting.”

Provided by PLOS

Women with Type 2 Diabetes More Likely to Develop Mental Health Conditions

Men, on the other hand, experience more cardiovascular and renal disease post-diagnosis

Photo by isens usa on Unsplash

In the years following type 2 diabetes diagnosis, women are more likely to develop mental health conditions, while men are more likely to develop cardiovascular and renal complications, according to a study published August 25th in the open access journal PLOS Medicine by Fabiola Eto from Queen Mary University of London, UK, and colleagues.

Type 2 diabetes, a metabolic disorder defined by high blood sugar, results from resistance to or insufficient production of insulin. Globally, type 2 diabetes affects about 536.6 million people between 20 and 79 years old, or 10.5% of the population; by 2045, this number is expected to rise to 700 million. Cardiovascular disease, end-stage renal disease, and mental health disorders like depression and anxiety are all associated with type 2 diabetes. Women and men are known to experience these conditions differently, but research has not yet described the disease trajectories according to sex and age in people with type 2 diabetes.

Eto and colleagues examined anonymized health records from 28,720 men and women in the UK who developed type 2 diabetes between 2010 and 2020. Men constituted 62% of the cohort with a median age of 54 at type 2 diabetes onset; women were on average older at diagnosis.

In the years following diagnosis, 39% of the participants experienced at least one significant health event. The events’ trajectory and timing differed by sex. For example: women were more likely to develop mental health conditions post-diagnosis (8.1% compared to 5.3%) and to experience trajectories culminating in death, while men showed higher proportions of cardiovascular disease (8.4% compared to 5.3%), end-stage renal disease, and hypertension (21.1% compared to 20.2%). The researchers noted that on average, women used more health services and long-term prescriptions than men, which may contribute to the increased diagnoses.

There were several similarities between the sexes as well. Both women and men who had a combination of type 2 diabetes and a mental health condition experienced premature mortality compared to other trajectories. Across all age groups and sexes, hypertension was the most frequent event following type 2 diabetes.

Current UK medical guidelines lack sex-specific prevention strategies and management for type 2 diabetes and comorbidities, especially mental health conditions. This study highlights the need for healthcare interventions differentiated by characteristics like sex and age.

The authors add, “One of the clearest signals in our data was the sex difference. Younger women with type 2 diabetes were showing mental health complications earlier and more often than we expected, which suggests routine psychological screening should be incorporated into standard diabetes care. For men, the pattern looked different: cardiovascular and kidney risks tended to emerge earlier, pointing to a need for earlier monitoring and stronger strategies to support men’s engagement with their treatment over time.”

Provided by PLOS

Incorrect AI Advice is a Blind Spot – Even for Doctors

 New study highlights potential challenges for using automated tools in healthcare

Photo by Accuray on Unsplash

In experiments in which physicians made decisions about treating hypothetical patients, the physicians tended to trust incorrect advice presented as being generated by artificial intelligence (AI), even after given the opportunity to notice that patient recovery data contradicted the recommendations. Aranzazu Vinas of the University of the Basque Country, Spain, and colleagues present these findings in the open-access journal PLOS Digital Health. 

AI systems can help physicians categorise patients according to their different care needs, such as whether a patient is more or less likely to benefit from a certain treatment. Since these systems are not perfect, they are meant to be used as suggestions, with potential errors caught and corrected by physicians.

 Prior research has shown that, in general, people struggle to notice and correct mistakes made by AI. To explore how this challenge may extend to physicians, Vinas and colleagues analysed data from 223 physicians who anonymously participated in online experiments.

 The physicians were asked to imagine they had the option to treat patients for a rare disease using a not-yet-proven treatment still under development. They were told that an AI system had identified which patients were more or less likely to benefit from the treatment. The physicians then chose which patients to treat, and after being presented with data on patient recovery, rated their perceptions of how reliable the AI was.

 Crucially, the actual effectiveness of the hypothetical treatment did not align with the AI recommendations. In one experiment, the treatment was equally moderately effective for all patients, and in a second experiment, it was equally ineffective for all.

 However, in both experiments, the physicians tended to rate the AI system as reliable and apparently did not use the patient recovery data to conclude that the AI recommendations were incorrect. In the second experiment, the physicians did not realise that the treatment was entirely ineffective.

These findings highlight potential challenges for incorporating AI-based classification into healthcare. Future research could build on this study, such as by developing and testing strategies and protocols that could increase human critical thinking and detection of AI errors, in order to maximize the benefits of the human-AI collaboration while minimising potential errors.”

Lead author Aranzazu Vinas notes: ” In both experiments, physicians mostly trusted the AI’s classifications and had trouble learning from the feedback. Furthermore, in the second experiment, professionals did not notice that the treatment was completely ineffective.”

 Co-author Helena Matute adds, “People tend to say that there is always a human controlling the algorithm, but our experiments show that doctors (as well as anyone else) have problems in learning from the available evidence when it contradicts the suggestions of an algorithm.”

Co-author Fernando Blanco summarizes: “It is important to investigate the errors that humans (including doctors) make when working with algorithms, in order to learn how to minimize the problems that arise from them.”

Press Preview: https://plos.io/4wjPxSs

In your coverage please use this URL to provide access to the freely available article in PLOS Digital Health: https://plos.io/4blGKHA

Contact: Aranzazu Vinas, aranzazu.vinas@ehu.eus

Image Caption: Doctors working on an AI-support system

Image Credit: Photo by Accuray on Unsplash. Free to share under the Unsplash license.

High-Resolution Image Link: https://unsplash.com/photos/a-few-men-looking-at-a-computer-screen-S34fEzWT6eE

Proteins in Blood Could Help Predict Retinal Degeneration in Diabetics

A model using 71 proteins associated with retinal degradation could predict risk in diabetics

Plasma proteomic signatures for early risk stratification of diabetic retinal neurodegeneration.
Credit: Wei Wang and Huangdong Li / Zhongshan Ophthalmic Center, Sun Yat-sen University (CC-BY 4.0, https://creativecommons.org/licenses/by/4.0/)

An AI-assisted model based on 71 different blood proteins could help doctors better predict retinal degeneration in diabetic patients before symptoms occur, according to a study published June 2nd in the open access journal PLOS Medicine by Huangdong Li from the Guangdong Provincial Clinical Research Center for Ocular Diseases in Guangzhou, China, and colleagues.

More than half a billion people around the world are now affected by diabetes. People with the disease are at risk of different neurodegenerative conditions, including the breakdown of the retina, the part of the eye that detects light, in a condition called diabetic retinal neurodegeneration (DRN). It can cause severe visual impairment and vision loss, and scientists believe that DRN is a “window” into the diabetic degeneration of other parts of the nervous system, including cognitive impairment and dementia, as well as degradation of nerves in peripheral areas like the fingers and toes.

Unfortunately, DRN is only detected after symptoms appear, when damage is already irreversible. To better predict who might suffer from DRN and when, the researchers sampled the blood plasma from 1492 patients in the Guangzhou Diabetic Eye Study with type 2 diabetes who did not yet have DRN, and examined the eyes of 1218 of them through scans over a six-year period. They compared their results with another 502 people with diabetes in the United Kingdom BioBank.

The researchers identified 71 different plasma proteins associated with DRN. The proteins were part of cell pathways for processes like inflammation and cellular maintenance. Using machine learning, the scientists used the protein levels in plasma to develop a predictive model called Pro-DRN which was able to improve on the best-performing model by 26 percent. The scientists have already put the model online to allow doctors to assess the risk. While Pro-DRN is based on plasma protein levels and relies on associations between protein levels and DRN and not direct causes, the authors hope that it could help doctors predict and potentially prevent neurodegeneration, using a simple blood test analysed by AI.

The authors add, “Our study suggests that early retinal nerve damage in diabetes leaves measurable signals in the blood. By combining plasma proteomics, longitudinal retinal imaging, and explainable AI, Pro-DRN may help move diabetic eye care from detecting established damage toward earlier, molecularly informed risk stratification, so that closer monitoring and future neuroprotective interventions can be directed to the people most likely to benefit.”

Provided by PLOS