Tom Kelsey

STUDY QUESTION: Do premenarcheal girls have a high proportion of morphologically abnormal ovarian follicles? SUMMARY ANSWER: A higher prevalence of follicles with an abnormal number of oocytes and nuclei was found in premenarcheal ovarian tissue compared with postmenarcheal tissue. WHAT IS KNOWN ALREADY: Morphological aberrations of ovarian follicles, including multi-nuclear oocytes and multi-ovular follicles, have been documented across species, including humans. These specific morphological aberrations, together with other types of abnormal follicles, are more prevalent in the ovaries of children and decline in number with age. However, detailed studies of the prevalence of multi-nuclear and multi-ovular follicles in paediatric ovarian tissue are limited. STUDY DESIGN, SIZE, DURATION: This was an observational study of the ovaries of children and adolescents who underwent ovarian cryopreservation prior to chemotherapy. PARTICIPANTS/MATERIALS, SETTING, METHODS: The study recruited participants aged 0–18 years who underwent ovarian tissue cryopreservation at a single centre. Haematoxylin and eosin-stained slides of ovarian biopsies were scanned and analysed to determine follicle number and aberrations in morphology and the cortical area was measured. Abnormal follicles were defined as those with an abnormal number of oocytes or oocyte nuclei. Relationships with menarcheal status (premenarcheal; postmenarcheal) and age were assessed. MAIN RESULTS AND THE ROLE OF CHANCE: A total of 149 participants’ ovarian samples were evaluated in the study, of whom 67 (45%) had follicles with abnormal numbers of oocytes or oocyte nuclei. Amongst 117 participants who had adequate follicle numbers, these abnormal follicle types accounted for 3.8% (902/23 891) of the total number of follicles and were significantly more common in premenarcheal girls (52/80; 65.0%) compared with postmenarcheal girls (15/37; 40.5%; P = 0.013). Ovular abnormalities (in 76.7% of abnormal follicles) were more common than nuclear abnormalities (in 23.3% of abnormal follicles; P < 0.001). Bi-ovular primordial follicles were the most common abnormality, comprising 54% of the total abnormal follicle population and were present in 41.9% of participants. LIMITATIONS, REASONS FOR CAUTION: Small ovarian biopsies were analysed, which may not fully represent the whole ovary. There is also a limited number of participants in each menarcheal group. WIDER IMPLICATIONS OF THE FINDINGS: Our study highlights that although follicle numbers are high in the prepubertal ovary, there was also a high proportion of follicles containing abnormal numbers of oocytes and oocyte nuclei. Further longitudinal and functional studies are recommended to determine whether this impacts fertility and the longevity of function following ovarian tissue grafting for restoration of fertility in these participants. Our work supports a separate multioocytic follicle class within the NICHD follicle classification, and we suggest expanding it to include a distinct multinuclear follicle category (oocyte contains more than one nucleus), as these likely represent different aberrations in oogenesis or folliculogenesis. This classification with clear definitions and nomenclature may improve consistency in reporting and facilitate future research. FUNDING: This work was supported by an MRFF Grant (MRFAR000308) supporting the funding of coauthors M.J. and M.A. Y.J. is supported through a Dame Kate Campbell Fellowship, The University of Melbourne. Co-authors V.G.-L. and J.Y.M. are supported through National Institute of Health (NIH) Z1A HD009005. DISCLOSURES: No conflict of interests are declared. TRIAL REGISTRATION NUMBER: n/a.

Background: The last 40 years have seen a substantial improvement in overall survival from cancer in children and young people. There is limited information on psychiatric wellbeing in female survivors of cancer at a young age. Objective: In this 40-year population-based linkage study, we set out to determine the incidence of first psychiatric admission compared with a non-exposed age-matched control. Methods: Scottish cancer registry records from 1981 to 2012 were linked to psychiatric admissions, maternity and death records from January 1981 to September 2018 using the unique personal Community Health Index number allocated to each person in Scotland. For each exposed subject, three age and deprivation matched controls from the population were selected. The primary exposed group was all females with a cancer diagnosis at age <25 years and no previous pregnancy and no psychiatric admission before cancer diagnosis. The main outcome measure is admission to a psychiatric hospital with a mental health diagnosis. Findings: Female cancer survivors had a significantly lower cumulative incidence of first psychiatric admissions than matched controls over the 38 years of follow-up for the cohort (p<0.05). The relative risk of a first psychiatric admission at 25 years from cancer diagnosis was 0.72 (0.56–0.89). Conclusion: Overall, we have shown that young cancer survivors are less likely than age-matched controls to have a psychiatric admission after cancer diagnosis. In particular, psychiatric admissions for mood disorders, neuroses, personality disorders and substance use are significantly less likely in the cancer survivors. Clinical implications: The experience of cancer treatment and survival in young females may reduce the risk of psychiatric admission in later life.

Wi-Fi sensor networks have grown rapidly due to their scalability and high data throughput, finding applications in tasks like tracking human motion in laboratory settings. These systems detect motion by analyzing fluctuations in radio signals caused by target movements, which generate identifiable activity patterns. However, their performance is influenced by factors such as environmental changes, unseen target subjects, multi-target tracking, data configurations, and the nature of target activities. These challenges lead to domain shifts between the training and testing phases, a common issue in real-world scenarios known as the domain-shifting problem in transfer learning. We propose a supervised domain alignment technique to address domain shifts in Wi-Fi sensor Channel State Information (CSI) datasets using minimal labeled target data. Our method outperforms state-of-the-art adversarial models trained on similar data, achieving superior cross-domain prediction accuracy. Evaluations on two public CSI datasets show consistent improvements, with an average Micro-F1 score of 90% for cross-user tasks and 67% for cross-user and cross-environment tasks using only 70 labeled target samples. This approach demonstrates its effectiveness in enhancing prediction accuracy under challenging domain shift scenarios.

This article explores multilingual sentiment analysis of short texts using three commercial decoder-only Large Language Models (“LLMs” ): OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini. The training data for these models is approximately 90% English, and it remains an open question whether it is better to evaluate text data in its original language or translate it into English first. We build on previous research on sentiment analysis of multilingual short texts, such as those found on social media, using 1000 short text samples in seven languages (English, Spanish, French, Portuguese, Arabic, Japanese, and Korean) translated into English using Google Translate. We processed these samples with decoder-only LLMs and compared their results with those from other methods (encoder-only LLMs, RNNs, lexicons). We found that decoder-only LLMs achieved the highest accuracy across all sentiment analysis methods when working with the original language data. The only exception was with the French data, where an RNN was the most accurate. Among the three decoder-only LLMs, ChatGPT had the highest accuracy in four of the seven languages, Claude in two, and Gemini, which ranked second in six of the seven languages.

Background: Large language models (LLMs) have great potential to improve and make the work of clinicians more efficient. Previous studies have mainly focused on web-based services, such as ChatGPT, often with simulated cases. For the processing of personalized patient data, web-based services have major data protection concerns. Ensuring compliance with data protection and medical device regulations therefore remains a critical challenge for adopting LLMs in clinical settings. Objective: This retrospective single-center study aimed to evaluate locally run LLMs (Gemma 2, Mistral Nemo, and Llama 3) in providing diagnosis and treatment recommendation for real-world outpatient cases in otorhinolaryngology (ORL). Methods: Outpatient cases (n=30) from regular consultation hours and the emergency service at a university hospital ORL outpatient department were randomly selected. Documentation by ORL doctors, including anamnesis and examination results, was passed to the locally run LLMs (Gemma 2, Mistral Nemo, and Llama 3), which were asked to provide diagnostic and treatment strategies. Recommendations of the LLMs and the treating ORL doctors were rated by 3 experienced ORL consultants on a 6-point Likert scale for medical adequacy, conciseness, coherence, and comprehensibility. Moreover, consultants were asked whether the answers pose a risk to the patient’s safety. A modified Turing test was performed to distinguish responses generated by LLMs from those of doctors. Finally, the potential influence of the information generated by the LLMs on the raters’ own diagnosis and treatment opinions was evaluated. Results: Over all categories, ORL doctors achieved superior ( P <.0005) ratings compared to locally run LLMs (Llama 3, Mistral Nemo, and Gemma 2). ORL doctors’ responses were considered hazardous for patients in only 1% of the ratings, whereas recommendations by Llama 3, Gemma 2, and Mistral Nemo were considered hazardous in 54%, 47%, and 32% of cases, respectively. According to the raters, the LLM’s information rarely influenced their judgment, with Mistral Nemo, Gemma 2, and Llama 3 achieving 1%, 3%, and 4% of the ratings, respectively. Conclusions: Although locally run LLM models still underperform compared with their web-based counterparts, they achieved respectable results on outpatient treatment in this study. Nevertheless, the retrospective and single-center nature of the study, along with the clinicians’ documentation style, may have introduced bias in favor of human recommendations. In the future, locally run LLMs will help address data protection concerns; however, further refinement and prospective validation are still needed to meet strict medical device requirements. As locally run LLMs continue to evolve, they are likely to become comparably powerful to web-based LLMs and become established as useful tools to support doctors in clinical practice.

This paper investigates the changes in financial assets and markets from December 1st, 2021, to April 30th, 2022, during the start of the Ukraine War. These dates roughly correspond to the prelude to the War in December 2021 to a few weeks after Russian troops withdrew from the Kyiv area on April 7th, 2022. We used the Goldstein 1992 Results Table to create Positive and Negative Geopolitical Risk bigrams (Goldstein, 1992). With these bigrams, we collected over 3.6 million tweets during our research period in seven different languages (English, Spanish, French, Portuguese, Arabic, Japanese, and Korean) to capture worldwide reaction to the Ukraine War. Using various sentiment analysis methods, we constructed a time series of changes in the daily Geopolitical Risk sentiment. We explored its relationship to 39 financial assets and markets at various time lags. We found through Granger causality that the geopolitical risk time series contained predictive information on several assets and market changes.

Human–computer interaction (HCI) drives innovation by bridging humans and technology, with human activity recognition (HAR) playing a key role. Traditional HAR systems require user cooperation and infrastructure, raising privacy concerns. In recent years, Wi-Fi devices have leveraged channel state information (CSI) to decode human movements without additional infrastructure, preserving privacy. However, these systems struggle with unseen users, new environments, and scalability, thereby limiting real-world applications. Recent research has also demonstrated that the impact of surroundings causes dissimilar variations in the channel state information at different times of the day. In this paper, we propose an unsupervised multi-source domain adaptation technique that addresses these challenges. By aligning diverse data distributions with target domain variations (e.g., new users, environments, or atmospheric conditions), the method enhances system adaptability by leveraging public datasets with varying domain samples. Experiments on three public CSI datasets using a preprocessing module to convert CSI into image-like formats demonstrate significant improvements to baseline methods with an average micro-F1 score of 81% for cross-user, 76% for cross-user and cross-environment, and 73% for cross-atmospheric tasks. The approach proves effective for scalable, device-free sensing in realistic cross-domain HAR scenarios.

STUDY QUESTION: Does maternal age impact hormonal secretions from granulosa cells, theca cells, and the oocyte in human small antral follicles? SUMMARY ANSWER: Major hormones secreted by granulosa and theca cells, as well as the oocyte-specific TGF-β members—GDF9, BMP15, and the GDF9/BMP15 heterodimer cumulin—maintain a consistent concentration within the follicular fluid of human small antral follicles, regardless of maternal age. WHAT IS KNOWN ALREADY: It is well established that female fertility declines with increasing age. However, it is not known whether this decline is exclusively due to a reduction in oocyte quality and quantity or also involves a decline in the hormone-secreting capabilities of granulosa cells, theca cells, and the oocyte itself. STUDY DESIGN, SIZE, DURATION: This is a retrospective study of follicular fluid obtained from human small antral follicles collected in connection with cryopreservation of ovarian tissue at the Laboratory of Reproductive Biology, University Hospital Copenhagen, Rigshospitalet, Denmark, between 2010 and 2020 as part of the hospital’s fertility preservation program. PARTICIPANTS/MATERIALS, SETTING, METHODS: Follicular fluid samples from human small antral follicles measuring 3–13 mm in diameter from macroscopically normal ovaries of 381 patients aged 5–43 years were included in the study, provided that at least one of the following parameters was measured: AMH, Inhibin A, Inhibin B, oestradiol (E2), progesterone (P4), androstenedione, testosterone, and/or the oocyte-specific TGF-β members GDF9, BMP15, or cumulin. MAIN RESULTS AND THE ROLE OF CHANCE: In a linear regression analysis adjusted for follicular volume, female age did not predict the follicular fluid concentrations of AMH, Inhibin B, Inhibin A, E2, androstenedione, testosterone, GDF9, BMP15, or cumulin. Although a significant association was observed between female age and follicular fluid P4 levels, the predictive value of age was poor, accounting for at most 5% of the variation in P4. LIMITATIONS, REASONS FOR CAUTION: Hormonal levels may vary with the degree of atresia in each follicle; however, the health status of the small antral follicles in this study was not characterized. Additionally, we cannot exclude possible age-related differences in human follicles larger than 10 mm, as very few of these were included. Furthermore, we did not include women above the age of 43, despite the potential for more pronounced age-related effects in these patients. WIDER IMPLICATIONS OF THE FINDINGS: Our results support the idea that the age-related decline in female fertility is primarily due to a reduction in oocyte quality and quantity, but further research is needed to confirm this. STUDY FUNDING/COMPETING INTEREST(S): No specific funding was obtained, and the authors have no conflicts of interest to declare in relation to this work. TRIAL REGISTRATION NUMBER: N/A.

Infertility affects one-in-six couples, often necessitating in vitro fertilization treatment (IVF). IVF generates complex data, which can challenge the utilization of the full richness of data during decision-making, leading to reliance on simple ‘rules-of-thumb’. Machine learning techniques are well-suited to analyzing complex data to provide data-driven recommendations to improve decision-making. In this multi-center study (n = 19,082 treatment-naive female patients), including 11 European IVF centers, we harnessed explainable artificial intelligence to identify follicle sizes that contribute most to relevant downstream clinical outcomes. We found that intermediately-sized follicles were most important to the number of mature oocytes subsequently retrieved. Maximizing this proportion of follicles by the end of ovarian stimulation was associated with improved live birth rates. Our data suggests that larger mean follicle sizes, especially those >18 mm, were associated with premature progesterone elevation by the end of ovarian stimulation and a negative impact on live birth rates with fresh embryo transfer. These data highlight the potential of computer technologies to aid in the personalization of IVF to optimize clinical outcomes pending future prospective validation.

Hypertensive patients need regular checkups and constant monitoring for taking time critical decisions by the medical experts. Unfortunately, it is hard to maintain uninterrupted patient health surveillance due to limited medical staff resulting in an increasing mortality rate annually. Thanks to recent developments in wireless sensor networking, we can monitor constantly and efficiently diverse parameters of a network. Similarly, Wireless Body Area Networks (WBANs) have become a well‐known sub‐branch of Wireless Sensor Networks. Such sensor networks can be leveraged for patient health monitoring, minimising the medical staff workload. Wireless Body Area Networks require tiny sensor nodes with limited battery power. Therefore, it is always desirable to design effective routing schemes that can enhance network lifetime, and reduce packet drop ratio. In this paper, we re‐simulate and explain in detail the results of a selected published journal article for WBANs and provide some modifications to improve the network's overall performance. Based on these amendments, the modified protocol successfully extends the operational time of the network than the original. Our performance evaluation parameters are dead nodes, throughput, residual energy, and path loss versus the number of rounds. These analyses support effective solutions that improve network performance and data delivery ratio.