Adversarial AI applied to cross-user inter-domain and intra-domain adaptation in human activity recognition using wireless signals

In recent years, researchers have successfully recognised human activities using commercially available WiFi (Wireless Fidelity) devices. The channel state information (CSI) can be gathered at the access point with the help of a network interface controller (NIC card). These CSI streams are sensitive to human body motions and produce abrupt changes (fluctuations) in their magnitude and phase values when a moving object interacts with a transmitter and receiver pair. This sensing methodology is gaining popularity compared to traditional approaches involving wearable technology, as it is a contactless sensing strategy with no cumbersome sensing equipments fitted on the target with preserved privacy since no personal information of the subject is collected. In previous investigations, internal validation statistics have been promising. However, external validation results have been poor, due to model application to varying subjects with remarkably different environments. To address this problem, we propose an adversarial Artificial Intelligence AI model that learns and utilises domain-invariant features. We analyse model results in terms of suitability for inter-domain and intra-domain alignment techniques, to identify which is better at robustly matching the source to target domain, and hence improve recognition accuracy in cross-user conditions for HAR using wireless signals. We evaluate our model performance on different target training data percentages to assess model reliability on data scarcity. After extensive evaluation, our architecture shows improved predictive performance across target training data proportions when compared to a non-adversarial model for nine cross-user conditions with comparatively less simulation time. We conclude that inter-domain alignment is preferable for HAR applications using wireless signals, and confirm that the dataset used is suitable for investigations of this type. Our architecture can form the basis of future studies using other datasets and/or investigating combined cross-environmental and cross-user features.

Mini-AFTERc: a controlled pilot trial of a nurse-led psychological intervention for fear of breast cancer recurrence

OBJECTIVES

To determine the feasibility and acceptability of implementing the Mini-AFTERc intervention.

DESIGN

Non-randomised cluster-controlled pilot trial.

SETTING

Four NHS out-patient breast cancer centres in Scotland.

PARTICIPANTS

Ninety-two women who had successfully completed primary treatment for breast cancer were screened for moderate levels of fear of cancer recurrence (FCR). Forty-five were eligible (17 intervention and 28 control) and 34 completed 3-month follow-up (15 intervention and 21 control).

INTERVENTION

Mini-AFTERc, a single brief (30 min) structured telephone discussion with a specialist breast cancer nurse (SBCN) trained to target the antecedents of FCR.

OUTCOMES

Feasibility and acceptability of Mini-AFTERc and the study design were assessed via recruitment, consent, retention rates, patient outcomes (measured at baseline, 2, 4, and 12 weeks), and post-study interviews with participants and SBCNs, which were guided by Normalisation Process Theory.

RESULTS

Mini-AFTERc was acceptable to patients and SBCNs. SBCNs believe the implementation of Mini-AFTERc to be feasible and an extension of discussions that already happen routinely. SBCNs believe delivery, however, at the scale required would be challenging given current competing demands for their time. Recruitment was impacted by variability in the follow-up practices of cancer centres and COVID-19 lockdown. Consent and follow-up procedures worked well, and retention rates were high.

CONCLUSIONS

The study provided invaluable information about the potential challenges and solutions for testing the Mini-AFTERc intervention more widely where limiting high FCR levels is an important goal following recovery from primary breast cancer treatment.

TRIAL REGISTRATION

ClinicalTrials.gov, NCT0376382 . Registered on 4 December 2018.

Improved survival of children and adolescents with classical Hodgkin lymphoma treated on a harmonised protocol in South Africa

BACKGROUND

Historic South African 5-year overall survival (OS) rates for Hodgkin lymphoma (HL) from 2000 to 2010 were 46% and 84% for human immunodeficiency virus (HIV)-positive and HIV-negative children, respectively. We investigated whether a harmonised treatment protocol using risk stratification and response-adapted therapy could increase the OS of childhood and adolescent HL.

METHODS

Seventeen units prospectively enrolled patients less than 18 years, newly diagnosed with classical HL onto a risk-stratified, response-adapted treatment protocol from July 2016 to December 2022. Low- and intermediate-risk patients received four and six courses of adriamycin, bleomycin, vinblastine, and dacarbazine (ABVD), respectively. High-risk patients received two courses of ABVD, followed by four courses of cyclophosphamide, vincristine, prednisone, and dacarbazine (COPDac). Those with a slow early response and bulky disease received consolidation radiotherapy. HIV-positive patients could receive granulocyte colony-stimulating factor and less intensive therapy if stratified as high risk, at the treating clinician's discretion. Kaplan-Meier survival analysis was performed to determine 2-year OS and Cox regression to elucidate prognostic factors.

RESULTS

The cohort comprised 132 patients (19 HIV-positive, 113 HIV-negative), median age of 9.7 years, with a median follow-up of 2.2 years. Risk grouping comprised nine (7%) low risk, 36 (27%) intermediate risk and 87 (66%) high risk, with 71 (54%) rapid early responders and 45 (34%) slow early responders, and 16 (12%) undocumented. Two-year OS was 100% for low-risk, 93% for intermediate-risk, and 91% for high-risk patients. OS for HIV-negative (93%) and HIV-positive (89%) patients were similar (p = .53). Absolute lymphocyte count greater than 0.6 × 109 predicted survival (94% vs. 83%, p = .02).

CONCLUSION

In the first South African harmonised HL treatment protocol, risk stratification correlated with prognosis. Two-year OS of HIV-positive and HIV-negative patients improved since 2010, partially ascribed to standardised treatment and increased supportive care. This improved survival strengthens the harmonisation movement and gives hope that South Africa will achieve the WHO Global Initiative for Childhood Cancer goals.

ChatGPT Versus Consultants: Blinded Evaluation on Answering Otorhinolaryngology Case-Based Questions

BACKGROUND

Large language models (LLMs), such as ChatGPT (Open AI), are increasingly used in medicine and supplement standard search engines as information sources. This leads to more "consultations" of LLMs about personal medical symptoms.

OBJECTIVE

This study aims to evaluate ChatGPT's performance in answering clinical case-based questions in otorhinolaryngology (ORL) in comparison to ORL consultants' answers.

METHODS

We used 41 case-based questions from established ORL study books and past German state examinations for doctors. The questions were answered by both ORL consultants and ChatGPT 3. ORL consultants rated all responses, except their own, on medical adequacy, conciseness, coherence, and comprehensibility using a 6-point Likert scale. They also identified (in a blinded setting) if the answer was created by an ORL consultant or ChatGPT. Additionally, the character count was compared. Due to the rapidly evolving pace of technology, a comparison between responses generated by ChatGPT 3 and ChatGPT 4 was included to give an insight into the evolving potential of LLMs.

RESULTS

Ratings in all categories were significantly higher for ORL consultants (P<.001). Although inferior to the scores of the ORL consultants, ChatGPT's scores were relatively higher in semantic categories (conciseness, coherence, and comprehensibility) compared to medical adequacy. ORL consultants identified ChatGPT as the source correctly in 98.4% (121/123) of cases. ChatGPT's answers had a significantly higher character count compared to ORL consultants (P<.001). Comparison between responses generated by ChatGPT 3 and ChatGPT 4 showed a slight improvement in medical accuracy as well as a better coherence of the answers provided. Contrarily, neither the conciseness (P=.06) nor the comprehensibility (P=.08) improved significantly despite the significant increase in the mean amount of characters by 52.5% (n= (1470-964)/964; P<.001).

CONCLUSIONS

While ChatGPT provided longer answers to medical problems, medical adequacy and conciseness were significantly lower compared to ORL consultants' answers. LLMs have potential as augmentative tools for medical care, but their "consultation" for medical problems carries a high risk of misinformation as their high semantic quality may mask contextual deficits.

Models and Biomarkers for Ovarian Ageing

The human ovarian reserve is defined by the number of non-growing follicles (NGFs) in the ovary, with the age-related decline in NGF population determining age at menopause for healthy women. In this chapter, the concept of ovarian reserve is explored in detail, with a sequence of models described that in principle allow any individual to be compared to the general population. As there is no current technology that can count the NGFs in a living ovary, we move our focus to biomarkers for the ovarian reserve. Using serum analysis and ultrasound it is possible to measure anti-Müllerian hormone (AMH), follicle-stimulating hormone (FSH), and ovarian volume (OV) and to count numbers of antral follicles (AFC). These are compared, with ovarian volume being the closest to a true biomarker for a wide range of ages and with AMH and AFC being the most popular for post-pubertal and pre-menopausal ages. The study of genetic and subcellular biomarkers for the ovarian reserve has produced less concrete results. Recent advances are described and compared in terms of limitations and potential. The chapter concludes with an overview of the future study indicated by our current knowledge and by current controversy in the field.

O-177 Long-term follow up to assess criteria for ovarian tissue cryopreservation for fertility preservation in young women and girls with cancer

Study question

Do the Edinburgh selection criteria correctly identify females, diagnosed with cancer under 18 years old, at high risk of future premature ovarian insufficiency (POI)?

Summary answer

Patient assessment using these criteria accurately identifies those at risk of POI. Ovarian tissue cryopreservation with future transplantation can be offered, providing future fertility options.

What is known already

Cancer treatments can be gonadotoxic and future fertility and reproductive health should be considered at the time of diagnosis and treatment planning. Correct identification of patients at high risk allows appropriate discussion of fertility preservation with ovarian tissue cryopreservation (OTC) and future transplantation. The Edinburgh selection criteria have been proposed as a tool to identify those patients at high risk.

However, the surgical procedure is not without risk and reproductive outcomes remain uncertain in girls. Therefore, long-term follow up of reproductive function is crucial to ensure that this treatment strategy is offered appropriately.

Study design, size, duration

All females diagnosed with cancer less than 18 years old, in South East Scotland, between 01/01/96 and 30/10/20 were included. They were assessed using the Edinburgh selection criteria and offered OTC, if appropriate. Ongoing long-term follow up of reproductive outcomes has been undertaken for the whole patient cohort to detect those who develop POI.

Participants/materials, setting, methods

A total of 639 eligible patients were identified from the Cancer registry and their electronic records reviewed. Reproductive function was assessed by the presence of menstruation, pregnancy, hormonal measurements, evidence of puberty or diagnosis of POI. Patients on hormonal contraception (other than for the treatment of POI) were considered unsuitable for analysis.

Data were analysed using the Kaplan Meier method, with POI as the event, and the Cox proportional hazards model to calculate hazard ratios.

Main results and the role of chance

Of the 639 patients diagnosed with cancer during the study period, those deceased before age 12 years old (n = 73) or under 12 years old (n = 134) at the date of analysis were excluded; also excluding those on hormonal contraception (n = 9) gave a study population of 423.

Data were analysed including those with unknown reproductive outcomes (n = 143), assuming they did not have POI. A subgroup analysis excluding these patients was also performed.

Mean age at diagnosis and analysis was 8.8 years and 22.5 years respectively.

OTC was offered to 37 patients, 26 of whom underwent the procedure. Nine patients developed POI (24.3%). Of the 386 not offered OTC, 11 developed POI (2.85%). The hazard ratio for developing POI was 8.8 (CI 3.6-21).

Excluding the patients with unknown outcomes (n = 143) left a study population of 280. Within this group, 9 of 29 offered OTC developed POI (31.0%) versus 11 of 251 not offered OTC (4.4%); hazard ratio 8.2 (CI 3.4-20).

In the group offered OTC, all cases of POI developed after the primary treatment. In those not offered OTC, POI developed after secondary treatment for disease relapse in 5 patients (45.5%).

Limitations, reasons for caution

A significant number of patients had unknown reproductive outcomes; this is likely to reflect a lack of recording of normal menstrual function in oncology/haematology clinics but may have biased the analysis. The duration of follow up is also short for some patients, highlighting the need for further follow up.

Wider implications of the findings

The overall prevalence of POI after childhood cancer is low, but the Edinburgh selection criteria are a robust tool for selecting those at high risk at the time of diagnosis, who can be offered OTC. However, many patients had incomplete information on current reproductive status, which should be assessed routinely.

Trial registration number

N/A

OR20-06 Kisspeptin as a Biomarker for Pregnancy Complications

Background:

Placentation (invasion of the placenta into the maternal endometrium) is hypothesised to be critical for healthy placental function and is abnormal in two thirds of miscarriage. Kisspeptin has emerged as a putative regulator of physiological placentation; it is highly expressed in placental syncitio-trophoblasts, whereas its receptor is expressed in both syncitio- and cyto-trophoblasts, such that kisspeptin is hypothesized to play an important paracrine role to regulate placentation. Circulating kisspeptin levels are considerably raised during healthy pregnancy and are reduced in women with miscarriage.

Aim:

We aimed to investigate the utility of circulating kisspeptin concentrations in the assessment of pregnancy complications and assess whether kisspeptin provides additional diagnostic information compared to beta human chorionic gonadotropin (βhCG) alone.

Methods:

This study was performed in collaboration with the Early Pregnancy Outcome Study (EPOS), which aims to identify novel pregnancy biomarkers. Women were invited to attend every fortnight for blood-sampling, clinical and ultrasound assessment during the first trimester, and repeated during the second and third trimesters. Asymptomatic women with healthy pregnancy (n=265) provided 960 blood-samples. Women with pregnancy complications including miscarriage (n=95), pre-eclampsia (PET; n=24), pregnancy induced hypertension (PIH; n=14), gestational diabetes (GDM; n=41), preterm birth (PTB; n=14) and intrauterine growth restriction (IUGR; n=24) provided 569 blood-samples.

Results:

Gestation-adjusted circulating kisspeptin and βhCG levels were lower, by 66% and 57%, respectively, in women with miscarriage compared to healthy pregnant controls (p<0.0001). Area under ROC curve for diagnosis of miscarriage was greater for the combination of both kisspeptin and βhCG together (0.92) than for either measure alone (βhCG 0.859, kisspeptin 0.874). An adjusted logistic regression model revealed that an 100pmol/L increase in plasma kisspeptin reduced the odds of miscarriage by 42%. Gestation-adjusted kisspeptin levels were lower in women with GDM (P=0.002), or IUGR (P<0.0001), and higher in women with PTB (P=0.004). Kisspeptin increased with gestation greater in PET (P=0.008) and PIH (P<0.0001) than in healthy controls.

Conclusions:

Plasma kisspeptin is a promising biomarker for pregnancy complications and provides additional diagnostic capability over that provided by βhCG alone.

A controlled pilot trial of a nurse-led intervention (Mini-AFTERc) to manage fear of cancer recurrence in patients affected by breast cancer

BACKGROUND

Fear of cancer recurrence (FCR) is common in people affected by breast cancer. FCR is associated with increased health service and medication use, anxiety, depression and reduced quality of life. Existing interventions for FCR are time and resource intensive, making implementation in a National Health Service (NHS) setting challenging. To effectively manage FCR in current clinical practice, less intensive FCR interventions are required. Mini-AFTERc is a structured 30-min counselling intervention delivered over the telephone and is designed to normalise moderate FCR levels by targeting unhelpful behaviours and misconceptions about cancer recurrence.This multi-centre non-randomised controlled pilot trial will investigate the feasibility of delivering the Mini-AFTERc intervention, its acceptability and usefulness, in relation to specialist breast cancer nurses (SBCNs) and patients. This protocol describes the rationale, methods and analysis plan for this pilot trial of the Mini-AFTERc intervention in everyday practice.

METHODS

This study will run in four breast cancer centres in NHS Scotland, two intervention and two control centres. SBCNs at intervention centres will be trained to deliver the Mini-AFTERc intervention. Female patients who have completed primary breast cancer treatment in the previous 6 months will be screened for moderate FCR (FCR4 score: 10‑14). Participants at intervention centres will receive the Mini-AFTERc intervention within 2 weeks of recruitment. SBCNs will audio record the intervention telephone discussions with participants. Fidelity of intervention implementation will be assessed from audio recordings. All participants will complete three separate follow-up questionnaires assessing changes in FCR, anxiety, depression and quality of life over 3 months. Normalisation process theory (NPT) will form the framework for semi-structured interviews with 20% of patients and all SBCNs. Interviews will explore participants' experience of the study, acceptability and usefulness of the intervention and factors influencing implementation within clinical practice. The ADePT process will be adopted to systematically problem solve and refine the trial design.

DISCUSSION

Findings will provide evidence for the potential effectiveness, fidelity, acceptability and practicality of the Mini-AFTERc intervention, and will inform the design and development of a large randomised controlled trial (RCT).

TRIAL REGISTRATION

ClinicalTrials.gov: NCT0376382. Registered 4th December 2018, https://clinicaltrials.gov/ct2/show/NCT03763825.

Efficient identification of patients eligible for clinical studies using case-based reasoning on Scottish Health Research register (SHARE)

BACKGROUND

Trials often struggle to achieve their target sample size with only half doing so. Some researchers have turned to Electronic Health Records (EHRs), seeking a more efficient way of recruitment. The Scottish Health Research Register (SHARE) obtained patients' consent for their EHRs to be used as a searching base from which researchers can find potential participants. However, due to the fact that EHR data is not complete, sufficient or accurate, a database search strategy may not generate the best case-finding result. The current study aims to evaluate the performance of a case-based reasoning method in identifying participants for population-based clinical studies recruiting through SHARE, and assess the difference between its resultant cohort and the original one deriving from searching EHRs.

METHODS

A case-based reasoning framework was applied to 119 participants in nine projects using two-fold cross-validation, with records from a further 86,292 individuals used for testing. A prediction score for study participation was derived from the diagnosis, procedure, pharmaceutical prescription, and laboratory test results attributes of each participant. Evaluation was conducted by calculating Area Under the ROC Curve and information retrieval metrics for the ranking list of the test set by prediction score. We compared the most likely participants as identified by searching a database to those ranked highest by our model.

RESULTS

The average ROCAUC for nine projects was 81% indicating strong predictive ability for these data. However, the derived ranking lists showed lower predictive performance, with only 21% of the persons ranked within top 50 positions being the same as identified by searching databases.

CONCLUSIONS

Case-based reasoning is may be more effective than a database search strategy for participant identification for clinical studies using population EHRs. The lower performance of ranking lists derived from case-based reasoning means that patients identified as highly suitable for study participation may still not be recruited. This suggests that further study is needed into improvements in the collection and curation of population EHRs, such as use of free text data to aid reliable identification of people more likely to be recruited to clinical trials.

Endocrine Requirements for Oocyte Maturation Following hCG, GnRH Agonist, and Kisspeptin During IVF Treatment

OBJECTIVE

The maturation of oocytes to acquire competence for fertilization is critical to the success of in vitro fertilization (IVF) treatment. It requires LH-like exposure, provided by either human chorionic gonadotropin (hCG), or gonadotropin releasing hormone agonist (GnRHa). More recently, the hypothalamic stimulator, kisspeptin, was used to mature oocytes. Herein, we examine the relationship between the endocrine changes following these agents and oocyte maturation.

DESIGN

Retrospective cohort study.

METHODS

Prospectively collected hormonal data from 499 research IVF cycles triggered with either hCG, GnRHa, or kisspeptin were evaluated.

RESULTS

HCG-levels (121 iU/L) peaked at 24 h following hCG, whereas LH-levels peaked at ~4 h following GnRHa (140 iU/L), or kisspeptin (41 iU/L). HCG-levels were negatively associated with body-weight, whereas LH rises following GnRHa and kisspeptin were positively predicted by pre-trigger LH values. The odds of achieving the median mature oocyte yield for each trigger were increased by hCG/LH level. Progesterone rise during oocyte maturation occurred precipitously following each trigger and strongly predicted the number of mature oocytes retrieved. Progesterone rise was positively associated with the hCG-level following hCG trigger, but negatively with LH rise following all three triggers. The rise in progesterone per mature oocyte at 12 h was greater following GnRHa than following hCG or kisspeptin triggers.

CONCLUSION

The endocrine response during oocyte maturation significantly differed by each trigger. Counter-intuitively, progesterone rise during oocyte maturation was negatively associated with LH rise, even when accounting for the number of mature oocytes retrieved. These data expand our understanding of the endocrine changes during oocyte maturation and inform the design of future precision-triggering protocols.

Anti-Müllerian hormone (AMH) in the Diagnosis of Menstrual Disturbance Due to Polycystic Ovarian Syndrome

Introduction: Polycystic ovarian syndrome (PCOS) is a leading cause of female subfertility worldwide, however due to the heterogeneity of the disorder, the criteria for diagnosis remains subject to conjecture. In the present study, we evaluate the utility of serum Anti-Müllerian hormone (AMH) in the diagnosis of menstrual disturbance due to PCOS. Method: Menstrual cycle length, serum AMH, gonadotropin and sex-hormone levels, total antral follicle count (AFC), body mass index (BMI) and ovarian morphology on ultrasound were analyzed in a cohort of 187 non-obese women, aged 18-35 years, screened for participation in a clinical trial of fertility treatment between 2013 and 2016 at a tertiary reproductive endocrine center. Results: Serum AMH was higher in women with menstrual disturbance when compared to those with regular cycles (65.6 vs. 34.8 pmol/L; P < 0.0001). The odds of menstrual disturbance was increased 28.5-fold (95% CI 3.6-227.3) in women with serum AMH >60 pmol/L, in comparison to those with an AMH < 15 pmol/L. AMH better discriminated women with menstrual disturbance (area under ROC 0.77) from those with regular menstrual cycles than AFC (area under ROC 0.67), however the combination of the two markers increased discrimination than either measure alone (0.83; 95% CI 0.77-0.89). Serum AMH was higher in women with all three cardinal features of PCOS (menstrual disturbance, hyperandrogenism, polycystic ovarian morphology) when compared to women with none of these features (65.6 vs. 14.6 pmol/L; P < 0.0001). The odds of menstrual disturbance were increased by 10.7-fold (95% CI 2.4-47.1) in women with bilateral polycystic morphology ovaries than those with normal ovarian morphology. BMI was a stronger predictor of free androgen index (FAI) than either AMH or AFC. Conclusion: Serum AMH could serve as a useful biomarker to indicate the risk of menstrual disturbance due to PCOS. Women with higher AMH levels had increased rates of menstrual disturbance and an increased number of features of PCOS.

Development, acceptability and feasibility of a communication skills training package for therapeutic radiographers to reduce fear of recurrence development in breast cancer patients (FORECAST2)

BACKGROUND

Many patients who have been treated for breast cancer experience high levels of fear that the cancer will return. The FORECAST pilot study showed that for a third of the patients, fears of cancer recurrence (FCR) increase during radiotherapy treatment and that conversations with their therapeutic radiographer at the weekly review meetings might help patients manage these concerns. This study aims to develop a communication skills training package (KEW, for 'Know', 'Encourage' and 'Warm-up') for therapeutic radiographers based on the findings of the FORECAST pilot study and on active input from patients and radiographers. This package will be piloted in a single centre to evaluate its acceptability and to prepare for a multi-centre clinical trial.

METHODS

The study consists of three phases. In the first phase, patient representatives and therapeutic radiographers participate in Experience-Based Co-Design to identify ways to improve communication during the radiotherapy review. In the second phase, various stakeholders, including members of the Society of Radiographers and of national patient representation groups, are consulted to develop a storyboard for the production of the communication training package. In the third phase, the acceptability and feasibility of the training is evaluated through observations, recruitment rates and follow-up discussions; a fidelity measure is designed; and potential benefits are observed, with patients' fear of cancer recurrence (FCR7) as the primary outcome. Secondary outcomes include a short daily measure of recurrence (FCR3), patients' positive and negative affect (PANAS), perceived empathy from the radiographer (CARE), satisfaction with the review meetings (RISS) and health-related quality of life (EQ-5D-3L).

DISCUSSION

To date, there has been limited research on how communication between therapeutic radiographers and patients during review appointments can help to manage patients' recurrence fears during radiotherapy treatment. A collaborative and participatory approach to the development of a communication skills training will ensure that it is optimally targeted to the needs and preferences of both patients and radiographers. Targeting recurrence fears through communication at this stage, when patients are still in regular contact with healthcare providers, has the potential to reduce the need for complex interventions post-treatment.

TRIAL REGISTRATION

NRES reference: 18/LO/0669. Clinicaltrial.gov ID: NCT03468881.