Warming up recurrent neural networks to maximise reachable multistability greatly improves learning

Training recurrent neural networks is known to be difficult when time dependencies become long. In this work, we show that most standard cells only have one stable equilibrium at initialisation, and that learning on tasks with long time dependencies generally occurs once the number of network stable equilibria increases; a property known as multistability. Multistability is often not easily attained by initially monostable networks, making learning of long time dependencies between inputs and outputs difficult. This insight leads to the design of a novel way to initialise any recurrent cell connectivity through a procedure called "warmup" to improve its capability to learn arbitrarily long time dependencies. This initialisation procedure is designed to maximise network reachable multistability, i.e., the number of equilibria within the network that can be reached through relevant input trajectories, in few gradient steps. We show on several information restitution, sequence classification, and reinforcement learning benchmarks that warming up greatly improves learning speed and performance, for multiple recurrent cells, but sometimes impedes precision. We therefore introduce a double-layer architecture initialised with a partial warmup that is shown to greatly improve learning of long time dependencies while maintaining high levels of precision. This approach provides a general framework for improving learning abilities of any recurrent cell when long time dependencies are present. We also show empirically that other initialisation and pretraining procedures from the literature implicitly foster reachable multistability of recurrent cells.

Parallax Inference for Robust Temporal Monocular Depth Estimation in Unstructured Environments

Estimating the distance to objects is crucial for autonomous vehicles, but cost, weight or power constraints sometimes prevent the use of dedicated depth sensors. In this case, the distance has to be estimated from on-board mounted RGB cameras, which is a complex task especially for environments such as natural outdoor landscapes. In this paper, we present a new depth estimation method suitable for use in such landscapes. First, we establish a bijective relationship between depth and the visual parallax of two consecutive frames and show how to exploit it to perform motion-invariant pixel-wise depth estimation. Then, we detail our architecture which is based on a pyramidal convolutional neural network where each level refines an input parallax map estimate by using two customized cost volumes. We use these cost volumes to leverage the visual spatio-temporal constraints imposed by motion and make the network robust for varied scenes. We benchmarked our approach both in test and generalization modes on public datasets featuring synthetic camera trajectories recorded in a wide variety of outdoor scenes. Results show that our network outperforms the state of the art on these datasets, while also performing well on a standard depth estimation benchmark.

Health workforce needs of small medical specialties: findings from rheumatology in Germany

Background

Small medical specialties may be more vulnerable to workforce shortage and the COVID-19 pandemic and this may directly impact in the provision of care for chronically-ill patients. This study aims to explore health workforce development and new needs, using rheumatology in Germany as a case study.

Methods

An explorative multi-methods approach was applied, combining health labour market assessment of rheumatology physicians (public statistics 2000-2019) and a questionnaire-based online survey conducted in early 2021 (n = 101 respondents; rheumatology physicians and residents). Main selected topics: work hours, workload, mental health issues, discrimination and sexual harassment experiences, impact of COVID-19. Descriptive statistical analysis was performed and qualitative content analysis for free-text information.

Results

Health labour market analysis showed that the numbers of rheumatologists increased markedly between 2000 and 2019 in the groups aged +50 years, but only 9% in younger groups under 50 years; since 2010 the group 40-50 years showed decreases. In 2019, the absolute number of rheumatologists working in healthcare after retirement-age exceeded those aged 40 and under. Survey data revealed a strong mismatch between actual and desired work hours for women and men. 81% rated their workload as high or very high; every sixth rheumatologist has suffered from stress or burnout syndromes at least once in the past. Experiences of gender discrimination and sexual harassment/violence were frequently reported, mostly by women. COVID-19 was an amplifier of stress with major stressors being digitalisation and increased demand for communication and patient education.

Conclusions

Decreasing health workforce capacities in German rheumatology combine with negative perceptions of work and workplace conditions, threatening both retention and service delivery.

Key messages

• Small medical specialties, like rheumatology, face severe shortage that threaten healthcare for chronically-ill patients and need greater attention.

• COVID-19 has reinforced rheumatologists’ workload and stressors, thus worsing mental health and retention.

POS0789 CLINICAL FEATURES OF SJÖGREN’S SYNDROME WITH AND WITHOUT NEUROLOGICAL INVOLVEMENT (NEURO-SJÖGREN)

Background

Sjögren’s Syndrome is well known for its characteristic sicca symptoms due to autoinflammatory destruction of the salivary and lacrymal glands, but neurological involvement is also common in this entity. Nevertheless, previously published smaller studies suggested distinct clinical features for Sjögren’s syndrome with and without neurological involvement, such as a more balanced gender distribution and lower IgG levels in patients with Sjögren‘ syndrome and neurological involvement1,2.

Objectives

We therefore aimed to systematically assess clinical features of patients with Sjögren’s syndrome with and without neurological involvement, find relevant in-between group differences and hereby aid early detection of both patient groups in the clinical routine to facilitate further studies, potentially with new therapeutic approaches.

Methods

We retrospectively assessed patients with Sjögren’s syndrome treated at the neurological and rheumatological/ immunological departement of our university hospital between 05/2014 and 09/2021 for available laboratory and clinical data. The displayed data represent preliminary results of this ongoing study.

Results

405 patients, who fulfilled the current ACR/EULAR classification criteria for Sjögren’s syndrome3 were currently included in the study (median age 59years [IQR 50-70 years], median ESSDAI 10 [IQR 3-16]). 228 patients (56%) showed neurological involvement. They were significantly more often male (32% vs. 14%; p<0.001) and showed lower IgG serum levels (median 11 g/l [IQR 9-13 g/l] vs 12 g/l [IQR 10-16 g/l], p<0.01) in comparison to patients with Sjögren’s syndrome but without neurological involvement. However, presence of objective xerostomia, objective xerophthalmia, SSA(Ro)-antibody-positivity or sialadenits grade 3 or 4 (Chisholm and Mason) on salivary gland biopsy did not differ between the two groups.

Conclusion

Preliminary analysis of this ongoing study supports the hypothesis, that patients with Sjögren’s syndrome and neurological impairment might express a distinct clinical phenotype in comparison to patients with Sjögren’s syndrome but without neurological involvement.

References

[1]Sjögren’s syndrome should be considered in patients with motor neuropathy. Zeitschrift fur Rheumatologie 2020; 79: 707–709.

[2]Seeliger T, Prenzler NK, Gingele S, et al. Neuro-Sjögren: Peripheral neuropathy with limb weakness in Sjögren’s syndrome. Front Immunol 2019; 10: 1600.

[3]Shiboski CH, Shiboski SC, Seror R, et al. 2016 American College of Rheumatology/European League Against Rheumatism classification criteria for primary Sjögren’s syndrome. Ann Rheum Dis 2017; 76: 9–16.

Disclosure of Interests

Tabea Seeliger: None declared, Emelie Kramer: None declared, Franz F. Konen: None declared, Sonja Beider: None declared, Alexandra Jablonka: None declared, Torsten Witte: None declared, Thomas Skripuletz Employee of: Honoraria for lectures: Alexion, Alnylam, Bayer Vital, Biogen, Celgene, CSL Behring, Euroimmun, Janssen, Merck Serono, Novartis, Pfizer, Roche, Sanofi, Siemens, Diana Ernst Consultant of: Participation in Advisory Boards: Abbvie, Galapagos, Amgen, Novartis, Employee of: Fees for Presentations: Abbvie, Amgen, BMS, Chugai, Cilag-Janssen, Galapagos, GSK, Medac, Lilly, Pfizer, Novartis, Roche

AB0536 RELATIONSHIP OF EXERTIONAL ACTIVITY AND MENTAL WELLBEING IN SYSTEMIC LUPUS ERYTHEMATOSUS

Background

Systemic lupus erythematosus (SLE) is a chronic autoimmune disease with a variety of clinical manifestations [1] and has a profound effect on physical health [2]. Anxiety and depression are common symptoms of SLE which can have a significant impact on the quality of life [3].

Objectives

To study whether or not the physical state of health relates to the extent of anxiety and depression in SLE patients.

Methods

The Physical Component Summary Score (PCS) was determined using the 36-Item Short Form Survey (SF-36) in a cohort of 146 SLE patients consecutively visiting our outpatient clinic. Two groups were compared: one group with low PCS<50 % (LPCS) and the comparison group with high PCS ≥50 % (HPCS). Patients with SLE who visited our Rheumatology clinic between March 2019 and December 2020 as part of a monocentric cross-sectional study completed additional standardized questionnaires: Hospital Anxiety and Depression Scale (HADS), Beck’s Depression Inventory (BDI II), Multidimensional Assessment of Fatigue (MAF), Functional questionnaire Hanover (FFbH) and the international questionnaire on physical activity in short form (IPAQ-SF). The data was analyzed with SPSS 27 (IBM, Armonk, NY, USA). The tests include bivariate and partial correlations. Significance tests were performed using non-parametric tests.

Results

In total, 146 patients participated in the study. 14.4 % (n = 21) were male and the mean age was 48.6 ± 12.2 years. The mean state of physical health according to SF36 was 54.9 % ± 25.6 % in the examined cohort.

There were 62 (42.5 %) patients with LPCS, of average age 50.4 ± 10.3 years and 84 (57.5 %) with HPCS, of average age 47.2 ± 13.3 years. The majority of patients in both groups were female (55/7, 88.7 % and 70/14, 83.3 %).

A SLEDAI score > 2 was confirmed in 33.9 % (n = 21) of the patients with LPCS and in 25 % (n = 21) of patients with HPCS, the mean SLEDAI score did not significantly differ between LPCS and HPCS patients (2.4 versus 1.9, p = 0.261). Among patients with LPCS, 54.8 % (n = 34) presented low physical activity and 64.5 % reported functional impairment (FFbH < 80). More than half of LPCS-patients (53.2 %, n = 33) showed a low Mental Health summary score (MCS < 50 %) and over a third of them had moderate to severe depression (BDI II > 19, 38.7 %, n = 24) and indicated pathological anxiety (HADS > 10, 37.1 %, n = 23).

Compared to the LPCS group, a smaller number of patients with HPCS had low physical activity (36.9 %, n = 31, p = 0.031) and the difference in mean weekly MET between the two groups was 2730 counts (p = 0.003). There was no impairment of functional capacity in patients with HPCS (FFbH mean Score 90.9 %, p < 0.001). In contrast to the LPCS group, only 17.9 % (n = 15) of the patients with HPCS showed MCS < 50 % (p < 0.001). Completely free of depressive symptoms were 65.5 % (n = 55) of the patients with HPCS and only 4 patients (4.8 %) reported moderate to severe depression (p < 0.001).

Patients with LPCS reported fatigue (GFI > 20) more often than patients with HPCS (98.4 % versus 54.8, p < 0.001).

Conclusion

SLE patients with low physical health conditions have highly significant mental health impairment, particularly anxiety and depression. Physical functioning and limitations due to physical health should be considered and physical activity needs to be improved. Measurement of the PCS should be a routine tool in the overall assessment of the health conditions of SLE patients.

References

[1]Lisnevskaia, L., G. Murphy, and D. Isenberg, Systemic lupus erythematosus. Lancet, 2014. 384(9957): p. 1878-1888.

[2]O’Dwyer, T., L. Durcan, and F. Wilson. Exercise and physical activity in systemic lupus erythematosus: A systematic review with meta-analyses. in Seminars in arthritis and rheumatism. 2017. Elsevier.

[3]Figueiredo-Braga, M., et al., Depression and anxiety in systemic lupus erythematosus: the crosstalk between immunological, clinical, and psychosocial factors. Medicine, 2018. 97(28).

Disclosure of Interests

None declared

POS0737 SICCA SYNDROME: A MULTIMODAL ASSESSEMENT TO CHARACTERIZE PATIENT COHORTS BEYOND SJÖGREN’S SYNDROME CLASSIFICATION CRITERIA

Background

Sicca syndrome represents a heterogeneous group of conditions, including Sjögren syndrome, causing xerophthalmia and xerostomia.

Objectives

This study characterizes in depth patients with Sicca syndrome and evaluates salivary gland ultrasound (SGUS) in this cohort.

Methods

Principal component analysis and hierarchical clustering of clinical parameters, including ESSPRI, ESSDAI and laboratory data were performed on all referrals for assessment of Sicca symptoms between October 2018 and March 2021. SGUS and labial gland biopsies were compared across groups.

Results

583 patients were assessed. Objective dryness was confirmed in 73% of patients. Cluster analysis identified 3 groups with post-hoc analysis confirming distinct phenotypes: Somatic Group (283/583; 49%) with higher reported symptoms but limited objective dryness; Dry Without Autoimmune Features (DAFneg, 206/532; 35%) and Dry With Autoimmune Features (DAFpos, 94/532; 16%). DAFpos patients had highest autoantibody titres (SSA 240 vs 3.6 vs 3.8; p<0.001), most extra-glandular manifestations (p<0.001) and highest median SGUS Score (DAFpos: 8 [IQR 4-10] vs SG: 2 [1-4] vs DAFneg 4 [2-5]; p< 0.001). No tangible correlation primary Sjögren syndrome criteria was observed.

Conclusion

SGUS score correlated with a subset of Sjögren syndrome patients, identified in the DAFpos cluster. This study highlights heterogeneity within Sicca and indeed Sjögren syndrome, highlighting the need for further studies.

References

[1]Chisholm DM, Mason DK. Labial salivary gland biopsy in Sjogren’s disease. J Clin Pathol. 1968;21(5):656-60

[2]De Vita S, Lorenzon G, Rossi G, Sabella M, Fossaluzza V. Salivary gland echography in primary and secondary Sjogren’s syndrome. Clin Exp Rheumatol. 1992;10(4):351-6

Table 1.

Comparing and contrasting the clinical demographics and attributes of the entire cohort subdivided into the three groups identified through principal component analysis and subsequent hierarchical clustering. Results are shown as mean and interquartile range unless otherwise stated.

SomaticDAFnegDAFposp N (%)283(49)206(35)94(16) Female, n (%)239(84)142(69)81(86%)<0.001 Age at Onset, yrs47.3[36.6-55.9]60.2[51.1-67.3]50.1[35.5-59.4]<0.001 BMI, kgm-226.1[23.0-31.0]24.7[21.7-27.7]24.6[22.4-28.0]0.003 Smoker, n (%)50(18)15(7)4(4)0.06ESSPRI Scores  - Dryness6[3-7]2[1-3]4[2-7]<0.001  - Limb Pain7[5-8]5[2-6]6[4-8]  - Fatigue8[6-9]3[2-5]5[3-8]Reported Symptoms Raynaud, n (%)86(30)51(25)44(47)0.006 Arthralgia, n (%)222(78)118(57)61(65)0.002 Myalgia, n (%)197(70)87(42)40(43)0.003 Stiffness, n (%)98(35)37(18)23(25)0.001 Parotitis, n (%)62(22)24(12)33(35)0.001 Sand corn, n (%)168(59)62(30)44(47)0.001 Ocular Inf, n (%)120(42)45(22)26(27)0.005ESSDAI  - Score5[2-12]5[0-11]11[4-17]<0.001Antibody Titres  - ANA ≥ 1:160178(63)152(74)44(47)<0.001  - RhF U/ml10.0[10.0-10.9]10.0[10.0-11.3]23.3[11.7-71.0]<0.001 - Alpha-Fodrin U/ml9[5-22]9[6-19]12[6-25]0.05  - anti-SSA(Ro) U/ml3.6[0.3-101.3]3.8[0.3-102.3]240.0[192.8-240.0]<0.001  - anti-SSB(La) U/ml0.4[0.3-3.4]0.3[0.3-1.9]73.1[3.8-312.5]<0.001Measurable Dryness Saxon, g3.5[2.4-4.9]4.2[3.3-5.3]2.3[0.6-3.7]<0.001 Schirmer, mm7.0[2.0-17.9]3.0[0.5-12.0]2.5[0.0-7.1]<0.001Labial Gland Biopsy, n (%) - Biopsy performed150(53)120(58)18(19) - Chisholm Score ≥366(44)64(53)9(50) - Median Score2[1-3]3[2-3]3[3-4]Salivary Gland Ultrasound, n (%) - SGUS = 039(14)39(19)1(1) - SGUS ≥673(26)55(27)38(41) SGUS Score2[1-4]4[2-5]8[4-10]<0.001Figure 1.

A 3D scatterplot composed of the first 3 dimensions of the principal component analysis, identified as providing greatest inertia gain on hierarchical clustering. The Somatic Group (SG) are represented by the black points. Patients in the Dryness without autoimmune features (DAFneg) are represented by pink points and Dryness with autoimmune features (DAFpos) patients with green dots.

Acknowledgements

We would like to express our gratitude to the staff of the Rheumatology Outpatients Department at Hannover Medical School for their continual help in organization of patients: G Mielke, A Lahn, Dr. S Hirsch.

Disclosure of Interests

Emelie Kramer: None declared, Tabea Seeliger: None declared, Thomas Skripuletz: None declared, Vega Gödecke: None declared, Sonja Beider: None declared, Alexandra Jablonka: None declared, Torsten Witte: None declared, Diana Ernst Grant/research support from: This study was financially supported by Novartis. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article or the decision to submit it for publication.

Jointly Learning Environments and Control Policies with Projected Stochastic Gradient Ascent

We consider the joint design and control of discrete-time stochastic dynamical systems over a finite time horizon. We formulate the problem as a multi-step optimization problem under uncertainty seeking to identify a system design and a control policy that jointly maximize the expected sum of rewards collected over the time horizon considered. The transition function, the reward function and the policy are all parametrized, assumed known and differentiable with respect to their parameters. We then introduce a deep reinforcement learning algorithm combining policy gradient methods with model-based optimization techniques to solve this problem. In essence, our algorithm iteratively approximates the gradient of the expected return via Monte-Carlo sampling and automatic differentiation and takes projected gradient ascent steps in the space of environment and policy parameters. This algorithm is referred to as Direct Environment and Policy Search (DEPS). We assess the performance of our algorithm in three environments concerned with the design and control of a mass-spring-damper system, a small-scale off-grid power system and a drone, respectively. In addition, our algorithm is benchmarked against a state-of-the-art deep reinforcement learning algorithm used to tackle joint design and control problems. We show that DEPS performs at least as well or better in all three environments, consistently yielding solutions with higher returns in fewer iterations. Finally, solutions produced by our algorithm are also compared with solutions produced by an algorithm that does not jointly optimize environment and policy parameters, highlighting the fact that higher returns can be achieved when joint optimization is performed.

Increasing planning targets is not enough: health workforce management in rheumatology in Germany

Background

Health workforce shortage in German rheumatology and its negative impact in healthcare are increasingly recognised. Health policy has increased staffing targets, but effective intervention strategies are lacking. This paper aims to systematically map the rheumatology workforce and to explore opportunities for intervention, thus contributing to better health workforce management.

Methods

The WHO National Health Workforce Accounts (NHWA) provided a conceptual framework for the mapping exercise. Four major sets of indicators were selected, comprising staffing levels, health labour market flows, skill-mix and education/training. Comparison of age-groups and time series was applied to explore trends. Public statistics and other secondary sources served the analysis, using descriptive methodology.

Results

Rheumatologists nearly doubled in absolute numbers since 2000, but this trend shows a problematic demographic bias. Between 2000 and 2019 numbers increased markedly in the groups aged +50 years, but only about 9% in younger groups under 50 years. Most alarming, since 2010 the group 40 to 50 years showed a reverse trend and strong decreases. In 2019, absolute numbers of rheumatologists working in healthcare after retirement-age exceeded those aged 40 and under. Since 2015 an expanding workforce trend overall flattened, but strongest in the hospital sector; annual inflows do not show relevant growths.

Conclusions

Health workforce assessment reveals negative demographic trends and decreasing appeal of the specialty especially to men, thus highlighting a need for more effective health workforce management. Intervention strategies should target innovation in resident training and task shifting and improve gender equality.

Key messages

Health workforce trends indicate that new health policy planning targets cannot be met with an available stock of rheumatologist. The WHO NHWA provide a useful toolbox to map health workforce trends and identify weaknesses and gaps.

A bio-inspired bistable recurrent cell allows for long-lasting memory

Recurrent neural networks (RNNs) provide state-of-the-art performances in a wide variety of tasks that require memory. These performances can often be achieved thanks to gated recurrent cells such as gated recurrent units (GRU) and long short-term memory (LSTM). Standard gated cells share a layer internal state to store information at the network level, and long term memory is shaped by network-wide recurrent connection weights. Biological neurons on the other hand are capable of holding information at the cellular level for an arbitrary long amount of time through a process called bistability. Through bistability, cells can stabilize to different stable states depending on their own past state and inputs, which permits the durable storing of past information in neuron state. In this work, we take inspiration from biological neuron bistability to embed RNNs with long-lasting memory at the cellular level. This leads to the introduction of a new bistable biologically-inspired recurrent cell that is shown to strongly improves RNN performance on time-series which require very long memory, despite using only cellular connections (all recurrent connections are from neurons to themselves, i.e. a neuron state is not influenced by the state of other neurons). Furthermore, equipping this cell with recurrent neuromodulation permits to link them to standard GRU cells, taking a step towards the biological plausibility of GRU. With this link, this work paves the way for studying more complex and biologically plausible neuromodulation schemes as gating mechanisms in RNNs.

AB0322 ASSOCIATION OF DEPRESSIVE SYMPTOMS ASSESSED BY BDI II WITH PHYSICAL ACTIVITY IN PATIENTS WITH PRIMARY SJÖGREN´S SYNDROME

Background:

Patients with primary Sjögren´s syndrome (pSS) suffer from pain, oral or ocular dryness and fatigue. Such symptoms can have a detrimental impact on health-related quality of life. The prevalence of mental health disorders in patients with pSS is considerably higher than in the general population [1]. Regular physical activity, such as nordic walking, improves aerobic capacity and may reduce reported fatigue [2]. According to current European League Against Rheumatism (EULAR) recommendations, management of pSS concomitant diseases should be evaluated in patients presenting with fatigue and pain, and severity scored using specific tools [3].

Objectives:

This study analyses the frequency and severity of depressive symptoms in patients with pSS and their relationship to fatigue and physical activity.

Methods:

In this monocentric, cross-sectional study, patients with pSS attending our Rheumatology clinic between January 2019 and March 2020 completed standardized questionnaires: Beck´s depression inventory second edition (BDI II) and the international physical activity questionnaire short form (IPAQ-SF). Data were analyzed using SPSS 26 (IBM, Armonk, NY, USA). The tests include bivariate and partial correlations and nonparametric Kruskal-Wallis-Test.

Results:

In total, 134 patients were included. The majority were female (117/134, 87.3 %), and the patients median age was 57 [21 – 85] years. Median duration of disease-related symptoms at inclusion was 56 months (range 0-388 months). Physical activity was low in 44. 8 % (n = 60), moderate in 32.1 % (n = 43), and high in 23.1 % (n = 31) of patients. Depressive symptoms of varying severity were identified in 76/134 (56.7%) of patients. Severe depression occurred in 10/76 (13.2%) and moderate depression in 14/76 (18.4%) patients. The remaining 52/76 (68.4%) patients had minimal or mild depression. Over two-thirds (67.7%) of patients reporting high physical activity showed no depression and no one with a high level of physical activity had severe depression. There was a clear association between the BDI II depression score and the level of physical activity (p = 0.003) regardless of age and illness duration. However, not all the symptoms assessed by BDI II had the same impact. From 21 items of the BDI II Inventory, symptoms such as loss of energy and fatigue were profoundly relevant, being reported by over 75 % of patients. The main depressive symptoms, which had a strong negative correlation with the level of physical activity were loss of energy (p < 0.001), sadness (p=0.018), inability to make decisions (p=0.007) and loss of pleasure (p=0.004). Somatic symptoms of decreased activation such as fatigue (p=0.026) and concentration difficulty (p=0.013) were also significant. The cognitive, self-negative symptoms were not associated with the level of physical activity.

Conclusion:

A negative correlation of physical activity with fatigue and depression in patients with pSS has been demonstrated. The somatic symptoms of decreased activation played a profound role in this association, while the cognitive self-negotiation symptoms were not influenced by physical activity in our patients. Such symptoms as fatigue and loss of energy could potentially be reduced by exercises and functional training and consequently decrease the level of depression. An intervention study in this regard would be recommended.

References:

[1]Cui, Y., et al., Anxiety and depression in primary Sjögren’s syndrome: a cross-sectional study. BMC psychiatry, 2018. 18(1): p. 1-8.

[2]Strömbeck, B., E. Theander, and L. Jacobsson, Effects of exercise on aerobic capacity and fatigue in women with primary Sjögren’s syndrome. Rheumatology, 2007. 46(5): p. 868-871.

[3]Ramos-Casals, M., et al., EULAR recommendations for the management of Sjögren’s syndrome with topical and systemic therapies. Annals of the rheumatic diseases, 2020. 79(1): p. 3-18.

Disclosure of Interests:

None declared

AB0707 RHEUMATOLOGY PATIENT CARE IN THE COVID-19 PANDEMIC: TELEMEDICINE, DELEGATION, PATIENT SATISFACTION AND VACCINATION BEHAVIOUR

Background:

Use of telemedicine in Germany has increased due to the COVID-19 lockdown. Between March and May 2020, government restrictions led to cancellation of routine outpatient appointments to limit viral spread and optimize resources.

Objectives:

This study assesses patient satisfaction of follow-up telemedicine appointments among patients known to be in disease remission, attending either secondary or tertiary care Rheumatology clinics. Appointments were conducted either by a rheumatologist or a qualified medical assistant for rheumatology (RFA). Additional data regarding perceived concerns arising from the COVID-19 pandemic as well as attitudes to vaccination were collected.

Methods:

Methods: Patients not requiring adjustment of their DMARDs at the two previous attendances were considered stable. At cancellation of the planned attendance, patients were offered participation in the study and provided verbal informed consent. Participants were randomized to a telemedicine appointment by either a physician or RFA. Telemedicine appointments consisted of a standardized patient interview, including assessment of disease activity (modified CDAI score), attitudes to vaccination as well as current vaccine status and concerns about COVID-19. Following participation, all patients received a pseudonymized postal questionnaire to evaluate appointment satisfaction (FAPI-Score).

Results:

In total 112/116 (96%) patients that were offered appointments, participated in the study (RA 50%, axSpA 30%, PsA 20%). Of these 88/112 (79%) returned their postal questionnaires. Overall patient satisfaction was excellent (mean 4.3/5 modified FAPI score) and did not differ between care setting or clinical status of the interviewer. RFAs conducted 19/112 (17%) of appointments, 6 (32%) of which required additional physician intervention. Change of DMARDs occurred in 19/112 (17%) appointments. Patients reporting a pain score ≥7 (VAS 1-10) were most dissatisfied with the telemedicine appointment (p=0.036). Concerns about COVID-19 correlated with disease activity: high disease activity (p = 0.031), presence of tender joints (p=0.001), high pain levels (p=0.009) correlated with concern of contracting COVID-19 or experiencing severe disease course. Only 38% of the patients had been vaccinated against pneumococci in the past 5 years and 54% had been vaccinated against influenza in 2019/2020.

Conclusion:

Telemedicine can contribute to patient care in stable patients. RFAs can also contribute to patient care especially for follow-up appointment when patients are in remission. Vaccination rates and motivation needs to be improved as influenza and pneumococcal vaccination is recommended to all patients with rheumatic diseases without contraindications.

Disclosure of Interests:

None declared

[Association of physical activity with fatigue and functional capacity in patients with rheumatoid arthritis]

BACKGROUND

Patients with rheumatoid arthritis (RA) tend to be less physically active. Physical activity has been shown to have a positive impact on disease activity and quality of life and is recommended by the European League Against Rheumatism (EULAR) as an integral component of standard treatment.

OBJECTIVE

A cross-sectional analysis of RA patients was carried out assessing disease activity, functional capacity and fatigue associated with physical activity.

MATERIAL AND METHODS

Physical activity, functional capacity and the global fatigue index (GFI) were examined using standardized questionnaires: the international physical activity questionnaire short form (IPAQ-SF), the functional questionnaire Hannover (FFbH) and the multidimensional assessment of fatigue (MAF). The data were evaluated using SPSS 26 (IBM, Armonk, NY, USA). The level of significance was tested with bivariate and partial correlations and nonparametric tests.

RESULTS

In total 164 patients were included in the study. The majority of the patients were female (127/164; 77%) and the median age of the cohort was 58.3 years (range 21-86 years). The median duration of disease-related symptoms at inclusion was 169 months (range 0-713 months). Physical activity was low in 39%, moderate in 37% and high in 24%. Patients reporting higher levels of physical activity reported significantly lower GFI (p < 0.001), functional limitations (p < 0.001) and disease activity (p = 0.045) scores than those with less physical activity.

CONCLUSION

Physical activity in RA patients was significantly correlated with functional capacity and levels of fatigue. In order to reduce the proportion of patients with low physical activity, the possibilities for functional training should be expanded and the patients should be encouraged to undertake sporting activities.

The impact of different COVID-19 containment measures on electricity consumption in Europe

As of March 13, 2020, the director general of the World Health Organization (WHO) considered Europe as the centre of the global COVID-19 outbreak. All countries within Europe had a confirmed case of COVID-19 by March 17. In response to the pandemic, different European countries took different approaches. This paper compares the impact of different containment measures taken by European countries in response to COVID-19 on their electricity consumption profiles. The comparisons are made for Spain, Italy, Belgium and the UK as countries with severe restrictions, and for the Netherlands and Sweden as countries with less restrictive measures. The results show that the consumption profiles reflect the difference in peoples' activities in different countries using various measures.

The specialised health workforce matters: how to adapt the WHO National Health Workforce Accounts

Background

The importance of a sustainable health workforce is increasingly recognised. However, the specialised heath workforce remains marginal in health workforce planning, policy and research. This paper aims to improve the integration of small specialised professional groups in health workforce governance and to critically assess existing tools, using Germany and rheumatology physicians as a case study.

Methods

A qualitative explorative approach was applied, which draws on a scoping review of the literature, public statistics and expert information. The National Health Workforce Accounts (NHWA) and the WHO 'AAAQ' framework served as guidance to assess workforce development in rheumatology in Germany.

Results

On average physician density is high in Germany, yet access to specialised care may be seriously delayed and services may not be accessible to all patients; e.g. currently only half of the patients with spondyloarthritis are treated by a rheumatologist which impacts quality of care. An estimated 850 to 1000 additional rheumatologists are needed to ensure early diagnoses and treatment and prevent long-term health damage. However, shortage will even reinforce in future: annual inflow of newly registered rheumatologists over recent years did not exceed 30 to 50 physicians and young rheumatologists more often prefer part-time work.

Conclusions

Specialised health professions with small numbers and long duration of training face specific challenges to create a sustainable workforce. This calls for (1) greater investment in education and training, (2) greater attention to qualitative indicators, such as assess and prevention measurements like early diagnosis, and (3) a dynamic health workforce approach to respond to both new technologies/ treatments and changing preferences of the health professionals, like part-time work.

Key messages

Integrating the specialised health workforce in health workforce governance is important to improve prevention, quality and access to care. NHWA may be aligned to qualitative indicators and prevention policies to respond more effectively to future population needs.

[Severe polyneuropathy in primary Sjögren's syndrome : Sjögren's syndrome should be considered in patients with motor neuropathy]

A 64-year-old male patient developed over a period of 20 years a peripheral neuropathy symmetrically affecting the upper and lower limbs. The histological examination of a sural nerve biopsy revealed a severe axonal neuropathy. Despite extensive laboratory investigations including immunological and metabolic tests the origin could not be identified. Finally, a Schirmer test revealed xerophthalmia. A subsequent salivary gland biopsy from the lower lip revealed a grade III lymphocytic inflammation according to Chisholm and Mason and confirmed the diagnosis of Sjögren's syndrome.

THU0263 LUNG INVOLVEMENT IN PRIMARY SJÖGREN SYNDROME – AN UNDER-DIAGNOSED ENTITY

Background:

Interstitial lung disease (ILD) represents a frequent extra-glandular manifestation of primary Sjögren’s Syndrome (pSS). Limited published data regarding phenotyping and treatment exists. Advances in managing specific ILD phenotypes have not been comprehensively explored in patients with coexisting pSS.

Objectives:

This retrospective study aimed to phenotype lung diseases occurring in a well-described pSS cohort and describe treatment course and outcomes.

Methods:

Between April 2018 and September 2019, all pSS patients attending our Outpatient clinic were screened for possible lung involvement. Clinical, laboratory and computer tomography (CT) findings were analysed. Patients were classified according to CT findings into 5 groups: usual interstitial pneumonia (UIP), non-specific interstitial pneumonia (NSIP), desquamative interstitial pneumonia (DIP), combined pulmonary fibrosis and emphysema (CPFE) and unspecific-ILD.

Results:

Lung involvement was confirmed in 24/240 patients (10%). Clinically manifest pSS occurred later in patients with ILD vs. non-ILD (53.2 [42.0-61.7]vs.62.3 [55.6-68.8] years; p=0.0016). The commonest phenotype was UIP n=10 (41%), followed by NSIP n=7 (29%), DIP n=2 (8%), CPFE n=2 (8%) and unspecific-ILD n=3 (13%). Forced vital capacity (FVC) and carbon monoxide diffusion capacity (DLCO) appeared lower in UIP and DIP, without reaching a significant difference. Treatment focused universally on intensified immunosuppression, with 12/24 patients (50%) receiving cyclophosphamide. No anti-fibrotic treatments were used. Follow-up was median 13.2 [7.9-72.3] months, during which 6/24 (25%) patients exhibited a further decline in FVC. No deaths occurred.

Conclusion:

Lung involvement in pSS is heterogeneous. Better phenotyping and tailored treatment may improve outcomes and requires further evaluation in larger prospective studies.

Disclosure of Interests:

None declared

Introducing neuromodulation in deep neural networks to learn adaptive behaviours

Animals excel at adapting their intentions, attention, and actions to the environment, making them remarkably efficient at interacting with a rich, unpredictable and ever-changing external world, a property that intelligent machines currently lack. Such an adaptation property relies heavily on cellular neuromodulation, the biological mechanism that dynamically controls intrinsic properties of neurons and their response to external stimuli in a context-dependent manner. In this paper, we take inspiration from cellular neuromodulation to construct a new deep neural network architecture that is specifically designed to learn adaptive behaviours. The network adaptation capabilities are tested on navigation benchmarks in a meta-reinforcement learning context and compared with state-of-the-art approaches. Results show that neuromodulation is capable of adapting an agent to different tasks and that neuromodulation-based approaches provide a promising way of improving adaptation of artificial systems.

On Overfitting and Asymptotic Bias in Batch Reinforcement Learning with Partial Observability

This paper provides an analysis of the tradeoff between asymptotic bias (suboptimality with unlimited data) and overfitting (additional suboptimality due to limited data) in the context of reinforcement learning with partial observability. Our theoretical analysis formally characterizes that while potentially increasing the asymptotic bias, a smaller state representation decreases the risk of overfitting. This analysis relies on expressing the quality of a state representation by bounding $L_1$ error terms of the associated belief states.  Theoretical results are empirically illustrated when the state representation is a truncated history of observations, both on synthetic POMDPs and on a large-scale POMDP in the context of smartgrids, with real-world data. Finally, similarly to known results in the fully observable setting, we also briefly discuss and empirically illustrate how using function approximators and adapting the discount factor may enhance the tradeoff between asymptotic bias and overfitting in the partially observable context.

Benchmarking for Bayesian Reinforcement Learning

In the Bayesian Reinforcement Learning (BRL) setting, agents try to maximise the collected rewards while interacting with their environment while using some prior knowledge that is accessed beforehand. Many BRL algorithms have already been proposed, but the benchmarks used to compare them are only relevant for specific cases. The paper addresses this problem, and provides a new BRL comparison methodology along with the corresponding open source library. In this methodology, a comparison criterion that measures the performance of algorithms on large sets of Markov Decision Processes (MDPs) drawn from some probability distributions is defined. In order to enable the comparison of non-anytime algorithms, our methodology also includes a detailed analysis of the computation time requirement of each algorithm. Our library is released with all source code and documentation: it includes three test problems, each of which has two different prior distributions, and seven state-of-the-art RL algorithms. Finally, our library is illustrated by comparing all the available algorithms and the results are discussed.

Pollen-derived nonallergenic substances enhance Th2-induced IgE production in B cells

BACKGROUND

B cells play a central role in IgE-mediated allergies. In damaged airway epithelium, they are exposed directly to aeroallergens. We aimed to assess whether direct exposure of B cells to pollen constituents affects allergic sensitization.

METHODS

B cells from murine splenocytes and from blood samples of healthy donors were incubated for 8 days under Th2-like conditions with aqueous ragweed pollen extracts (Amb-APE) or its constituents. Secreted total IgM, IgG, and IgE was quantified by ELISA. Additionally, birch, grass, or pine-pollen extracts were tested. The number of viable cells was evaluated by ATP measurements. B-cell proliferation was measured by CFSE staining. IgE class switch was analyzed by quantitation of class switch transcripts. In an OVA/Alum i.p.-sensitization mouse model, Amb-APE was intranasally instilled for 11 consecutive days.

RESULTS

Upon Th2 priming of murine B cells, ragweed pollen extract caused a dose-dependent increase in IgE production, while IgG and IgM were not affected. The low-molecular-weight fraction and phytoprostane E1 (PPE1) increased IgE production, while Amb a 1 did not. PPE1 enhanced IgE also in human memory B cells. Under Th1 conditions, Amb-APE did not influence immunoglobulin secretion. The IgE elevation was not ragweed specific. It correlated with proliferation of viable B cells, but not with IgE class switch. In vivo, Amb-APE increased total IgE and showed adjuvant activity in allergic airway inflammation.

CONCLUSIONS

Aqueous pollen extracts, the protein-free fraction of Amb-APE, and the pollen-contained substance PPE1 specifically enhance IgE production in Th2-primed B cells. Thus, pollen-derived nonallergenic substances might be responsible for B-cell-dependent aggravation of IgE-mediated allergies.

Pollen-derived adenosine is a necessary cofactor for ragweed allergy

BACKGROUND

Ragweed (Ambrosia artemisiifolia) is a strong elicitor of allergic airway inflammation with worldwide increasing prevalence. Various components of ragweed pollen are thought to play a role in the development of allergic responses. The aim of this study was to identify critical factors for allergenicity of ragweed pollen in a physiological model of allergic airway inflammation.

METHODS

Aqueous ragweed pollen extract, the low molecular weight fraction or the major allergen Amb a 1 was instilled intranasally on 1-11 consecutive days, and allergic airway inflammation was evaluated by bronchoalveolar lavage, lung histology, serology, gene expression in lung tissue, and measurement of lung function. Pollen-derived adenosine was removed from the extract enzymatically to analyze its role in ragweed-induced allergy. Migration of human neutrophils and eosinophils toward supernatants of ragweed-stimulated bronchial epithelial cells was analyzed.

RESULTS

Instillation of ragweed pollen extract, but not of the major allergen or the low molecular weight fraction, induced specific IgG1 , pulmonary infiltration with inflammatory cells, a Th2-associated cytokine signature in pulmonary tissue, and impaired lung function. Adenosine aggravated ragweed-induced allergic lung inflammation. In vitro, human neutrophils and eosinophils migrated toward supernatants of bronchial epithelial cells stimulated with ragweed extract only if adenosine was present.

CONCLUSIONS

Pollen-derived adenosine is a critical factor in ragweed-pollen-induced allergic airway inflammation. Future studies aim at therapeutic strategies to control these allergen-independent pathways.

Mathematical Modeling of HIV Dynamics After Antiretroviral Therapy Initiation: A Review

This review shows the potential ground-breaking impact that mathematical tools may have in the analysis and the understanding of the HIV dynamics. In the first part, early diagnosis of immunological failure is inferred from the estimation of certain parameters of a mathematical model of the HIV infection dynamics. This method is supported by clinical research results from an original clinical trial: data just after 1 month following therapy initiation are used to carry out the model identification. The diagnosis is shown to be consistent with results from monitoring of the patients after 6 months. In the second part of this review, prospective research results are given for the design of individual anti-HIV treatments optimizing the recovery of the immune system and minimizing side effects. In this respect, two methods are discussed. The first one combines HIV population dynamics with pharmacokinetics and pharmacodynamics models to generate drug treatments using impulsive control systems. The second one is based on optimal control theory and uses a recently published differential equation to model the side effects produced by highly active antiretroviral therapy therapies. The main advantage of these revisited methods is that the drug treatment is computed directly in amounts of drugs, which is easier to interpret by physicians and patients.

Large vessel vasculitis and spondyloarthritis: coincidence or associated diseases?

INTRODUCTION

Although cardiac complications have been reported in established spondyloarthritis (SpA), little is known about peripheral axial SpA in large vessel vasculitis (LVV). The aim of this study was to assess the prevalence of SpA in patients with newly diagnosed LVV.

METHOD

Retrospective single-centre analysis of all newly diagnosed LVV patients was performed between January 2011 and December 2012. Vasculitides were confirmed on thoracic magnetic resonance imaging (MRI) or 18-fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT). Patients completed a standardized questionnaire incorporating the Berlin criteria to assess inflammatory back pain. Existing scans were reassessed for sacroiliitis and ferritin antibodies measured in all patients.

RESULTS

Fifteen patients exhibiting new LVV were identified. Diagnosis was confirmed using MRI in nine patients and FDG-PET/CT in six. Six patients (40%) fulfilled American College of Rheumatology (ACR) criteria for giant cell arteritis (GCA) and polymyalgia rheumatica (PMR), and nine PMR only. Four patients fulfilled the Berlin criteria for inflammatory back pain, with three demonstrating sacroiliitis on imaging. All remaining patients demonstrated no sacroiliitis. One further patient with LVV lacking features of inflammatory back pain had known psoriatic arthritis (PsA). Patients with coexisting SpA were younger (mean age 57 years vs. 66 years) and had higher C-reactive protein (CRP) levels (200 mg/L vs. 85 mg/L) at presentation. Four SpA patients and seven out of nine patients with isolated LVV had ferritin antibodies.

CONCLUSIONS

We have demonstrated a higher than anticipated prevalence of SpA in LVV, given the reported 0.5-1% prevalence in the general population. Coexisting SpA should be considered in LVV patients exhibiting inflammatory back pain despite steroid initiation. Ferritin antibodies demonstrated a similarly high prevalence in aortitis and SpA as reported previously in untreated GCA and PMR.

Batch Mode Reinforcement Learning based on the Synthesis of Artificial Trajectories

In this paper, we consider the batch mode reinforcement learning setting, where the central problem is to learn from a sample of trajectories a policy that satisfies or optimizes a performance criterion. We focus on the continuous state space case for which usual resolution schemes rely on function approximators either to represent the underlying control problem or to represent its value function. As an alternative to the use of function approximators, we rely on the synthesis of "artificial trajectories" from the given sample of trajectories, and show that this idea opens new avenues for designing and analyzing algorithms for batch mode reinforcement learning.

Outcomes of splenectomy in patients with common variable immunodeficiency (CVID): a survey of 45 patients

Splenectomy has been used in patients with common variable immunodeficiency disorders (CVID), mainly in the context of refractory autoimmune cytopenia and suspected lymphoma, but there are understandable concerns about the potential of compounding an existing immunodeficiency. With increasing use of rituximab as an alternative treatment for refractory autoimmune cytopenia, the role of splenectomy in CVID needs to be re-examined. This retrospective study provides the largest cohesive data set to date describing the outcome of splenectomy in 45 CVID patients in the past 40 years. Splenectomy proved to be an effective long-term treatment in 75% of CVID patients with autoimmune cytopenia, even in some cases when rituximab had failed. Splenectomy does not worsen mortality in CVID and adequate immunoglobulin replacement therapy appears to play a protective role in overwhelming post-splenectomy infections. Future trials comparing the effectiveness and safety of rituximab and splenectomy are needed to provide clearer guidance on the second-line management of autoimmune cytopenia in CVID.

Plasma deoxysphingolipids: a novel class of biomarkers for the metabolic syndrome?

AIMS/HYPOTHESIS

Sphingolipid synthesis is typically initiated by the conjugation of L-serine and palmitoyl-CoA, a reaction catalysed by serine palmitoyltransferase (SPT). SPT can also metabolise other acyl-CoAs (C(12) to C(18)) and other amino acids such as L-alanine and glycine, giving rise to a spectrum of atypical sphingolipids. Here, we aimed to identify changes in plasma levels of these atypical sphingolipids to explore their potential as biomarkers in the metabolic syndrome and diabetes.

METHODS

We compared the plasma profiles of ten sphingoid bases in healthy individuals with those of patients with the metabolic syndrome but not diabetes, and diabetic patients (n = 25 per group). The results were verified in a streptozotocin (STZ) rat model. Univariate and multivariate statistical analyses were used.

RESULTS

Deoxysphingolipids (dSLs) were significantly elevated (p = 5 × 10⁻⁶) in patients with the metabolic syndrome (0.11 ± 0.04 μmol/l) compared with controls (0.06 ± 0.02 μmol/l) but did not differ between the metabolic syndrome and diabetes groups. Levels of C(16)-sphingosine-based sphingolipids were significantly lowered in diabetic patients but not in patients with the metabolic syndrome but without diabetes (p = 0.008). Significantly elevated dSL levels were also found in the plasma and liver of STZ rats. A principal component analysis revealed a similar or even closer association of dSLs with diabetes and the metabolic syndrome in comparison with the established biomarkers.

CONCLUSIONS/INTERPRETATION

We showed that dSLs are significantly elevated in patients with type 2 diabetes mellitus and non-diabetic metabolic syndrome compared with healthy controls. They may, therefore, be useful novel biomarkers to improve risk prediction and therapy monitoring in these patients.

Forests under climate change and air pollution: gaps in understanding and future directions for research

Forests in Europe face significant changes in climate, which in interaction with air quality changes, may significantly affect forest productivity, stand composition and carbon sequestration in both vegetation and soils. Identified knowledge gaps and research needs include: (i) interaction between changes in air quality (trace gas concentrations), climate and other site factors on forest ecosystem response, (ii) significance of biotic processes in system response, (iii) tools for mechanistic and diagnostic understanding and upscaling, and (iv) the need for unifying modelling and empirical research for synthesis. This position paper highlights the above focuses, including the global dimension of air pollution as part of climate change and the need for knowledge transfer to enable reliable risk assessment. A new type of research site in forest ecosystems ("supersites") will be conducive to addressing these gaps by enabling integration of experimentation and modelling within the soil-plant-atmosphere interface, as well as further model development.

Incomplete filling of lithium heparin tubes affects the activity of creatine kinase and gamma-glutamyltransferase

This study aims to assess whether or not incomplete filling of primary lithium heparin tubes may influence the activity of creatine kinase (CK), alanine aminotransferase (ALT), aspartate aminotransferase (AST) and gamma-glutamyltransferase (GGT). Blood was drawn from 20 healthy volunteer using an identical sequence of tubes. First, a 6 mL, 13 x 100 mm 14 unit/mL lithium heparin Vacuette was filled and discharged. Then, three identical lithium heparin Vacuette tubes were filled, one to the nominal volume (i.e., full-draw tube), another with half of the nominal volume (half-draw tube) and the last with one-third of the nominal volume (low-draw tube). The plasma was separated and tested for CK (non-activated by N-acetylcysteine), AST and ALT on a Beckman Coulter Unicel DxC 800. Tests for CK were performed with a different reagent on a Beckman Coulter AU5800 (activated by N-acetylcysteine). Although the concentrations of ASL and ALT measured on the Unicel DxC and that of CK measured on the AU5800 did not change significantly across the different specimens, those of CK and GGT measured on the Unicel DxC 800 were significantly increased in the half-draw and low-draw tubes. The percentage bias of CK on the Unicel DxC 800 (using Bland Altman plots) was 3.3% and 7.9% for the half-draw and low-draw tubes, respectively, whereas that of GGT was 10.3% and 16.6% for the half-draw and low-draw tubes, respectively. These results suggest that short-draw lithium heparin tubes might be unsuitable for testing GGT and CK using specific combinations of reagents and instrumentation.