Investigating the fluctuating nature of post-COVID pain symptoms in previously hospitalized COVID-19 survivors: the LONG-COVID-EXP multicenter study

Objective

This cohort study used Sankey plots and exponential bar plots for visualizing the fluctuating nature and trajectory of post-COVID pain in previously hospitalized COVID-19 survivors.

Methods

A cohort of 1266 subjects hospitalised because of COVID-19 during the first wave of the pandemic were scheduled for a telephone interview at 8.4 (T1), 13.2 (T2), and 18.3 (T3) months in average after hospitalization for collecting data about post-COVID pain. Patients were asked for about pain symptomatology that was attributed to the infection. Hospitalization and clinical data were collected from medical records.

Results

The prevalence of myalgia as COVID-19-associated symptom was 29.82% (n = 389) at hospitalization (T0). The prevalence of post-COVID pain was 41.07% (n = 520) at T1, 34.29% (n = 434) at T2, and 28.47% (n = 360) at T3. The recovery exponential curve revealed a decrease trend visualizing that post-COVID pain improved over the time span investigated. Pain in the lower extremity and widespread pain were the most prevalent locations. Female sex (OR 1.507, 95% CI 1.047-2.169), pre-existing pain symptoms (OR 1.724, 95% CI 1.237-2.403), headache as onset-symptom (OR 2.374, 95% CI 1.550-3.639), days at hospital (OR 1.012, 95% CI 1.000-1.025), and presence of post-COVID pain at T1 (OR 13.243, 95% CI 9.428-18.601) were associated with post-COVID pain at T2. Only the presence of post-COVID pain at T1 (OR 5.383, 95% CI 3.896-7.439) was associated with post-COVID pain at T3.

Conclusion

Current results show a fluctuating evolution with a decreasing tendency of post-COVID pain during the first years after hospitalization. The development of post-COVID pain soon after SARS-CoV-2 infection predispose for long-lasting chronic pain.

Trajectory of anxiety/depressive symptoms and sleep quality in individuals who had been hospitalized by COVID-19: The LONG-COVID-EXP multicenter study

OBJECTIVE

To apply Sankey plots and exponential bar plots for visualizing the evolution of anxiety/depressive symptoms and poor sleep in previously hospitalized COVID-19 survivors.

METHODS

A sample of 1266 subjects who were hospitalized due to a SARS-CoV-2 from March-May 2020 were assessed at 8.4 (T1), 13.2 (T2) and 18.3 (T3) months after hospitalization. The Hospital Anxiety and Depression Scale was used to determine anxiety (HADS-A) and depressive (HADS-D) symptoms. The Pittsburgh Sleep Quality Index (PSQI) evaluated sleep quality. Clinical features, onset symptoms and hospital data were collected from medical records.

RESULTS

Sankey plots revealed that the prevalence of anxiety symptomatology (HADS-A ≥ 8 points) was 22.5% (n = 285) at T1, 17.6% (n = 223) at T2, and 7.9% (n = 100) at T3, whereas the prevalence of depressive symptoms (HADS-D ≥ 8 points) was 14.6% (n = 185) at T1, 10.9% (n = 138) at T2, and 6.1% (n = 78) at T3. Finally, the prevalence of poor sleep (PSQI≥8 points) decreased from 32.8% (n = 415) at T1, to 28.8% (n = 365) at T2, and to 24.8% (n = 314) at T3. The recovery curves show a decrease trend visualizing that these symptoms recovered the following years after discharge. The regression models did not reveal medical records associated with anxiety/depressive symptoms or poor sleep.

CONCLUSION

The use of Sankey plots shows a fluctuating evolution of anxiety/depressive symptoms and poor sleep during the first years after the infection. In addition, exponential bar plots revealed a decrease prevalence of these symptoms during the first years after hospital discharge. No risk factors were identified in this cohort.

Trajectory of post-COVID brain fog, memory loss, and concentration loss in previously hospitalized COVID-19 survivors: the LONG-COVID-EXP multicenter study

Objective

This study aimed to apply Sankey plots and exponential bar plots for visualizing the trajectory of post-COVID brain fog, memory loss, and concentration loss in a cohort of previously hospitalized COVID-19 survivors.

Methods

A sample of 1,266 previously hospitalized patients due to COVID-19 during the first wave of the pandemic were assessed at 8.4 (T1), 13.2 (T2), and 18.3 (T3) months after hospital discharge. They were asked about the presence of the following self-reported cognitive symptoms: brain fog (defined as self-perception of sluggish or fuzzy thinking), memory loss (defined as self-perception of unusual forgetfulness), and concentration loss (defined as self-perception of not being able to maintain attention). We asked about symptoms that individuals had not experienced previously, and they attributed them to the acute infection. Clinical and hospitalization data were collected from hospital medical records.

Results

The Sankey plots revealed that the prevalence of post-COVID brain fog was 8.37% (n = 106) at T1, 4.7% (n = 60) at T2, and 5.1% (n = 65) at T3, whereas the prevalence of post-COVID memory loss was 14.9% (n = 189) at T1, 11.4% (n = 145) at T2, and 12.12% (n = 154) at T3. Finally, the prevalence of post-COVID concentration loss decreased from 6.86% (n = 87) at T1, to 4.78% (n = 60) at T2, and to 2.63% (n = 33) at T3. The recovery exponential curves show a decreasing trend, indicating that these post-COVID cognitive symptoms recovered in the following years after discharge. The regression models did not reveal any medical record data associated with post-COVID brain fog, memory loss, or concentration loss in the long term.

Conclusion

The use of Sankey plots shows a fluctuating evolution of post-COVID brain fog, memory loss, or concentration loss during the first years after the infection. In addition, exponential bar plots revealed a decrease in the prevalence of these symptoms during the first years after hospital discharge. No risk factors were identified in this cohort.

From electronic health records to clinical management systems: how the digital transformation can support healthcare services

Healthcare systems worldwide are currently undergoing significant transformations in response to increasing costs, a shortage of healthcare professionals and the growing complexity of medical needs among the population. Value-based healthcare reimbursement systems are emerging as an attempt to incentivize patient-centricity and cost containment. From a technological perspective, the transition to digitalized services is intended to support these transformations. A Health Information System (HIS) is a technological solution designed to govern the data flow generated and consumed by healthcare professionals and administrative staff during the delivery of healthcare services. However, the exponential growth of digital capabilities and applied advanced analytics has expanded their traditional functionalities and brought the promise of automating administrative procedures and simple repetitive tasks, while enhancing the efficiency and outcomes of healthcare services by incorporating decision support tools for clinical management. The future of HIS is headed towards modular architectures that can facilitate implementation and adaptation to different environments and systems, as well as the integration of various tools, such as artificial intelligence (AI) models, in a seamless way. As an example, we present the experience and future developments of the European Clinical Database (EuCliD®). EuCliD is a multilingual HIS used by 20 000 nurses and physicians on a daily basis to manage 105 000 patients treated in 1100 clinics in 43 different countries. EuCliD encompasses patients' follow-up, automatic reporting and mobile applications while enabling efficient management of clinical processes. It is also designed to incorporate multiagent systems to automate repetitive tasks, AI modules and advanced dynamic dashboards.

Trajectory of Post-COVID Self-Reported Fatigue and Dyspnoea in Individuals Who Had Been Hospitalized by COVID-19: The LONG-COVID-EXP Multicenter Study

Fatigue and dyspnoea are common post-COVID symptoms. The aim of this study was to apply Sankey plots and exponential bar plots for visualizing the evolution and trajectory of post-COVID fatigue and dyspnoea symptoms in a cohort of previously hospitalized COVID-19 survivors. A total of 1266 previously hospitalized patients due to COVID-19 participated in this multicentre study. They were assessed at hospital admission (T0), 8.4 months (T1), 13.2 months (T2) and 18.3 months (T3) after hospital discharge and were asked about the presence of self-reported fatigue or dyspnoea symptoms. Fatigue was defined as a self-perceived feeling of constant tiredness and/or weakness whereas dyspnoea was defined as a self-perceived feeling of shortness of breath at rest. We specifically asked for fatigue and dyspnoea that participants attributed to the infection. Clinical/hospitalization data were collected from hospital medical records. The prevalence of post-COVID fatigue was 56.94% (n = 721) at T1, 52.31% (n = 662) at T2 and 42.66% (n = 540) at T3. The prevalence of dyspnoea at rest decreased from 28.71% (n = 363) at hospital admission (T0), to 21.29% (n = 270) at T1, to 13.96% (n = 177) at T2 and 12.04% (n = 153) at T3. The Sankey plots revealed that 469 (37.08%) and 153 (12.04%) patients exhibited fatigue and dyspnoea at all follow-up periods. The recovery exponential curves show a decreased prevalence trend, showing that fatigue and dyspnoea recover the following three years after hospitalization. The regression models revealed that the female sex and experiencing the symptoms (e.g., fatigue, dyspnoea) at T1 were factors associated with the presence of post-COVID fatigue or dyspnoea at T2 and T3. The use of Sankey plots shows a fluctuating evolution of post-COVID fatigue and dyspnoea during the first two years after infection. In addition, exponential bar plots revealed a decreased prevalence of these symptoms during the first years after. The female sex is a risk factor for the development of post-COVID fatigue and dyspnoea.

Prevalence of Self-Reported Anosmia and Ageusia in Elderly Patients Who Had Been Previously Hospitalized by SARS-CoV-2: The LONG-COVID-EXP Multicenter Study

We explored two different graph methods for visualizing the prevalence of self-reported post-COVID anosmia and ageusia in a large sample of individuals who had been previously hospitalized in five different hospitals. A cohort of 1266 previously hospitalized COVID-19 survivors participated. Participants were assessed at hospitalization (T0) and at three different follow-up periods: 8.4 (T1), 13.2 (T2), and 18.3 (T3) months after hospital discharge. They were asked about the presence of self-reported anosmia and ageusia that they attributed to infection. Anosmia was defined as a self-perceived feeling of complete loss of smell. Ageusia was defined as a self-perceived feeling of complete loss of taste. Data about hospitalization were recorded from medical records. The results revealed that the prevalence of anosmia decreased from 8.29% (n = 105) at hospitalization (T0), to 4.47% (n = 56) at T1, to 3.27% (n = 41) at T2, and 3.35% (n = 42) at T3. Similarly, the prevalence of ageusia was 7.10% (n = 89) at the onset of SARS-CoV-2 infection (T0), but decreased to 3.03% (n = 38) at T1, to 1.99% (n = 25) at T2, and 1.36% (n = 17) at T3. The Sankey plots showed that only 10 (0.8%) and 11 (0.88%) patients exhibited anosmia and ageusia throughout all the follow-ups. The exponential curves revealed a progressive decrease in prevalence, demonstrating that self-reported anosmia and ageusia improved in the years following hospitalization. The female sex (OR4.254, 95% CI 1.184-15.294) and sufferers of asthma (OR7.086, 95% CI 1.359-36.936) were factors associated with the development of anosmia at T2, whereas internal care unit admission was a protective factor (OR0.891, 95% CI 0.819-0.970) for developing anosmia at T2. The use of a graphical method, such as a Sankey plot, shows that post-COVID self-reported anosmia and ageusia exhibit fluctuations during the first years after SARS-CoV-2 infection. Additionally, self-reported anosmia and ageusia also show a decrease in prevalence during the first years after infection, as expressed by exponential bar plots. The female sex was associated with the development of post-COVID anosmia, but not ageusia, in our cohort of elderly patients previously hospitalized due to COVID-19.

Trajectory of Gastrointestinal Symptoms in Previously Hospitalized COVID-19 Survivors: The Long COVID Experience Multicenter Study

This multicenter cohort study used Sankey plots and exponential bar plots to visualize the fluctuating evolution and the trajectory of gastrointestinal symptoms in previously hospitalized COVID-19 survivors during the first 18 months after acute SARS-CoV-2 infection. A total of 1266 previously hospitalized COVID-19 survivors were assessed at four points: hospital admission (T0), at 8.4 months (T1), at 13.2 months (T2), and at 18.3 months (T3) after hospitalization. Participants were asked about their overall gastrointestinal symptoms and particularly diarrhea. Clinical and hospitalization data were collected from hospital medical records. The prevalence of overall gastrointestinal post-COVID symptomatology was 6.3% (n = 80) at T1, 3.99% (n = 50) at T2 and 2.39% (n = 32) at T3. The prevalence of diarrhea decreased from 10.69% (n = 135) at hospital admission (T0), to 2.55% (n = 32) at T1, to 1.04% (n = 14) at T2, and to 0.64% (n = 8) at T3. The Sankey plots revealed that just 20 (1.59%) and 4 (0.32%) patients exhibited overall gastrointestinal post-COVID symptoms or diarrhea, respectively, throughout the whole follow-up period. The recovery fitted exponential curves revealed a decreasing prevalence trend, showing that diarrhea and gastrointestinal symptoms recover during the first two or three years after COVID-19 in previously hospitalized COVID-19 survivors. The regression models did not reveal any symptoms to be associated with the presence of gastrointestinal post-COVID symptomatology or post-COVID diarrhea at hospital admission or at T1. The use of Sankey plots revealed the fluctuating evolution of gastrointestinal post-COVID symptoms during the first two years after infection. In addition, exponential bar plots revealed the decreased prevalence of gastrointestinal post-COVID symptomatology during the first three years after infection.

Toward Prediction of Financial Crashes with a D-Wave Quantum Annealer

The prediction of financial crashes in a complex financial network is known to be an NP-hard problem, which means that no known algorithm can efficiently find optimal solutions. We experimentally explore a novel approach to this problem by using a D-Wave quantum annealer, benchmarking its performance for attaining a financial equilibrium. To be specific, the equilibrium condition of a nonlinear financial model is embedded into a higher-order unconstrained binary optimization (HUBO) problem, which is then transformed into a spin-1/2 Hamiltonian with at most, two-qubit interactions. The problem is thus equivalent to finding the ground state of an interacting spin Hamiltonian, which can be approximated with a quantum annealer. The size of the simulation is mainly constrained by the necessity of a large number of physical qubits representing a logical qubit with the correct connectivity. Our experiment paves the way for the codification of this quantitative macroeconomics problem in quantum annealers.

Clustering analysis reveals different profiles associating long-term post-COVID symptoms, COVID-19 symptoms at hospital admission and previous medical co-morbidities in previously hospitalized COVID-19 survivors

PURPOSE

To identify subgroups of COVID-19 survivors exhibiting long-term post-COVID symptoms according to clinical/hospitalization data by using cluster analysis in order to foresee the illness progress and facilitate subsequent prognosis.

METHODS

Age, gender, height, weight, pre-existing medical comorbidities, Internal Care Unit (ICU) admission, days at hospital, and presence of COVID-19 symptoms at hospital admission were collected from hospital records in a sample of patients recovered from COVID-19 at five hospitals in Madrid (Spain). A predefined list of post-COVID symptoms was systematically assessed a mean of 8.4 months (SD 15.5) after hospital discharge. Anxiety/depressive levels and sleep quality were assessed with the Hospital Anxiety and Depression Scale and Pittsburgh Sleep Quality Index, respectively. Cluster analysis was used to identify groupings of COVID-19 patients without introducing any previous assumptions, yielding three different clusters associating post-COVID symptoms with acute COVID-19 symptoms at hospital admission.

RESULTS

Cluster 2 grouped subjects with lower prevalence of medical co-morbidities, lower number of COVID-19 symptoms at hospital admission, lower number of post-COVID symptoms, and almost no limitations with daily living activities when compared to the others. In contrast, individuals in cluster 0 and 1 exhibited higher number of pre-existing medical co-morbidities, higher number of COVID-19 symptoms at hospital admission, higher number of long-term post-COVID symptoms (particularly fatigue, dyspnea and pain), more limitations on daily living activities, higher anxiety and depressive levels, and worse sleep quality than those in cluster 2.

CONCLUSIONS

The identified subgrouping may reflect different mechanisms which should be considered in therapeutic interventions.

Clustering Analysis Identifies Two Subgroups of Women with Fibromyalgia with Different Psychological, Cognitive, Health-Related and Physical Features but Similar Widespread Pressure Pain Sensitivity

OBJECTIVE

Since identification of groups of patients can help to better understand risk factors related to each group and to improve personalized therapeutic strategies, this study aimed to identify subgroups (clusters) of women with fibromyalgia syndrome (FMS) according to pain-related, related-disability, neuro-physiological, cognitive, health-related, psychological or physical features.

METHODS

Demographic, pain-related, sensory-related, related-disability, psychological, health-related, cognitive, and physical variables were collected in 113 women with FMS. Widespread pressure pain thresholds (PPTs) were also assessed. K-means clustering was used to identify groups of women without any previous assumption.

RESULTS

Two clusters exhibiting similar widespread sensitivity to pressure pain (PPTs) but differing in the remaining variables were identified. Overall, women in one cluster exhibited higher pain intensity and related-disability, more sensitization-associated and neuropathic pain symptoms, higher kinesiophobia, hypervigilance and catastrophism levels, worse sleep quality, higher anxiety/depressive levels, lower health-related function, and worse physical function than women in the other cluster.

CONCLUSIONS

Cluster analysis identified one group of women with FMS exhibiting worse sensory, psychological, cognitive and health-related features. Widespread sensitivity to pressure pain seems to be a common feature of FMS. Current results suggest that this group of women with FMS may need to be treated differently.

Post-COVID-19 Symptoms 2 Years After SARS-CoV-2 Infection Among Hospitalized vs Nonhospitalized Patients

IMPORTANCE

Identification of long-term post-COVID-19 symptoms among hospitalized and nonhospitalized patients is needed.

OBJECTIVE

To compare the presence of post-COVID-19 symptoms 2 years after acute SARS-CoV-2 infection between hospitalized and nonhospitalized patients.

DESIGN, SETTING, AND PARTICIPANTS

A cross-sectional cohort study was conducted at 2 urban hospitals and general practitioner centers from March 20 to April 30, 2020, among 360 hospitalized patients and 308 nonhospitalized patients with acute SARS-CoV-2 infection during the first wave of the pandemic. Follow-up was conducted 2 years later.

MAIN OUTCOMES AND MEASURES

Participants were scheduled for a telephone interview 2 years after acute infection. The presence of post-COVID-19 symptoms was systematically assessed, with particular attention to symptoms starting after infection. Hospitalization and clinical data were collected from medical records. Between-group comparisons and multivariate logistic regressions were conducted.

RESULTS

A total of 360 hospitalized patients (162 women [45.0%]; mean [SD] age, 60.7 [16.1] years) and 308 nonhospitalized patients (183 women [59.4%]; mean [SD] age, 56.7 [14.7] years) were included. Dyspnea was more prevalent at the onset of illness among hospitalized than among nonhospitalized patients (112 [31.1%] vs 36 [11.7%]; P < .001), whereas anosmia was more prevalent among nonhospitalized than among hospitalized patients (66 [21.4%] vs 36 [10.0%]; P = .003). Hospitalized patients were assessed at a mean (SD) of 23.8 (0.6) months after hospital discharge, and nonhospitalized patients were assessed at a mean (SD) of 23.4 (0.7) months after the onset of symptoms. The number of patients who exhibited at least 1 post-COVID-19 symptom 2 years after infection was 215 (59.7%) among hospitalized patients and 208 (67.5%) among nonhospitalized patients (P = .01). Among hospitalized and nonhospitalized patients, fatigue (161 [44.7%] vs 147 [47.7%]), pain (129 [35.8%] vs 92 [29.9%]), and memory loss (72 [20.0%] vs 49 [15.9%]) were the most prevalent post-COVID-19 symptoms 2 years after SARS-CoV-2 infection. No significant differences in post-COVID-19 symptoms were observed between hospitalized and nonhospitalized patients. The number of preexisting medical comorbidities was associated with post-COVID-19 fatigue (odds ratio [OR], 1.93; 95% CI, 1.09-3.42; P = .02) and dyspnea (OR, 1.91; 95% CI, 1.04-3.48; P = .03) among hospitalized patients. The number of preexisting medical comorbidities (OR, 3.75; 95% CI, 1.67-8.42; P = .001) and the number of symptoms at the onset of illness (OR, 3.84; 95% CI, 1.33-11.05; P = .01) were associated with post-COVID-19 fatigue among nonhospitalized patients.

CONCLUSIONS AND RELEVANCE

This cross-sectional study suggested the presence of at least 1 post-COVID-19 symptom in 59.7% of hospitalized patients and 67.5% of nonhospitalized patients 2 years after infection. Small differences in symptoms at onset of COVID-19 were identified between hospitalized and nonhospitalized patients. Post-COVID-19 symptoms were similar between hospitalized and nonhospitalized patients; however, lack of inclusion of uninfected controls limits the ability to assess the association of SARS-CoV-2 infection with overall and specific post-COVID-19 symptoms 2 years after acute infection. Future studies should include uninfected control populations.

Differences in Long-COVID Symptoms between Vaccinated and Non-Vaccinated (BNT162b2 Vaccine) Hospitalized COVID-19 Survivors Infected with the Delta Variant

This study compared differences in the presence of post-COVID symptoms among vaccinated and non-vaccinated COVID-19 survivors requiring hospitalization due to the Delta (B.1.617.2) variant. This cohort study included hospitalized subjects who had survived SARS-CoV-2 infection (Delta variant) from July to August 2021 in an urban hospital in Madrid, Spain. Individuals were classified as vaccinated if they received full administration (i.e., two doses) of BNT162b2 (“Pfizer-BioNTech”) vaccines. Other vaccines were excluded. Those with just one dose of the BNT162b2 vaccine were considered as non-vaccinated. Patients were scheduled for a telephone interview at a follow-up around six months after infection for assessing the presence of post-COVID symptoms with particular attention to those symptoms starting after acute infection and hospitalization. Anxiety/depressive levels and sleep quality were likely assessed. Hospitalization and clinical data were collected from medical records. A total comprising 109 vaccinated and 92 non-vaccinated COVID-19 survivors was included. Vaccinated patients were older and presented a higher number of medical comorbidities, particular cardiorespiratory conditions, than non-vaccinated patients. No differences in COVID-19 onset symptoms at hospitalization and post-COVID symptoms six months after hospital discharge were found between vaccinated and non-vaccinated groups. No specific risk factor for any post-COVID symptom was identified in either group. This study observed that COVID-19 onset-associated symptoms and post-COVID symptoms six-months after hospitalization were similar between previously hospitalized COVID-19 survivors vaccinated and those non-vaccinated. Current data can be applied to the Delta variant and those vaccinated with BNT162b2 (Pfizer-BioNTech) vaccine.

Neural Network-Based Calculator for Rat Glomerular Filtration Rate

Glomerular filtration is a pivotal process of renal physiology, and its alterations are a central pathological event in acute kidney injury and chronic kidney disease. Creatinine clearance (ClCr), a standard method for glomerular filtration rate (GFR) measurement, requires a long and tedious procedure of timed (usually 24 h) urine collection. We have developed a neural network (NN)-based calculator of rat ClCr from plasma creatinine (pCr) and body weight. For this purpose, matched pCr, weight, and ClCr trios from our historical records on male Wistar rats were used. When evaluated on the training (1165 trios), validation (389), and test sets (660), the model committed an average prediction error of 0.196, 0.178, and 0.203 mL/min and had a correlation coefficient of 0.863, 0.902, and 0.856, respectively. More importantly, for all datasets, the NN seemed especially effective at comparing ClCr among groups within individual experiments, providing results that were often more congruent than those measured experimentally. ACLARA, a friendly interface for this calculator, has been made publicly available to ease and expedite experimental procedures and to enhance animal welfare in alignment with the 3Rs principles by avoiding unnecessary stressing metabolic caging for individual urine collection.

Symptoms Experienced at the Acute Phase of SARS-CoV-2 Infection as Risk Factor of Long-term Post-COVID Symptoms: The LONG-COVID-EXP-CM Multicenter Study

Objective

This multicenter study investigated clinical risk factors associated with the number of long-term symptoms after COVID.

Methods

Clinical features, symptoms at hospital admission, hospitalization data, and the number of symptoms after COVID was systematically assessed for patients who recovered from COVID-19 in 4 hospitals in Madrid (Spain) from February 20 to May 31, 2020.

Results

Overall, 1,969 patients (46.5% women, age: 61, SD: 16 years) were randomly assessed 8.4 months (SD 1.5) after hospital discharge. Female gender (odds ratio [OR] 1.82, 95% confidence interval [CI] 1.57-2.10), number of morbidities (OR 1.182, 95% CI 1.08-1.29), number of symptoms at hospital admission (OR 1.309, 95% CI 1.15-1.49) and days at the hospital (OR 1.01, 95% CI 1.007-1.017) were associated (all, p <0.001) with more long-term symptoms after COVID. Further, vomiting (OR 1.78, 95% CI 1.26-2.52), throat pain (OR 1.36, 95% CI 1.02-1.81), diarrhea (OR 1.51, 95% CI 1.25-1.82), dyspnea (OR 1.20, 95% CI 1.01-1.41), or headache (OR 1.50, 95% CI 1.28-1.75) as symptoms at hospital admission were also associated (all, p <0.01) with a higher number of symptoms after COVID.

Conclusion

This multicenter study found that a higher number of symptoms at hospital admission were the most relevant risk factor for developing more symptoms after COVID, supporting the assumption that a higher symptom load at the acute phase is associated with a greater likelihood of long-term symptoms after COVID.

How to assess the risks associated with the usage of a medical device based on predictive modeling: the case of an anemia control model certified as medical device

BACKGROUND

The successful application of Machine Learning (ML) to many clinical problems can lead to its implementation as a medical device (MD), which is important to assess the associated risks.

METHODS

An anemia control model (ACM), certified as MD, may face adverse events as a result of wrong predictions that are translated into suggestions of doses of erythropoietic stimulating agents to dialysis patients. Risks are assessed as the combination of severity and probability of a given hazard. While severities are typically assessed by clinicians, probabilities are tightly related to the performance of the predictive model.

RESULTS

A postmarketing data set formed by all adult patients registered in French, Portuguese, and Spanish clinics, belonging to an international network, was considered; 3876 patients and 11,508 suggestions were eventually included. The achieved results show that there are no statistical differences between the probabilities of adverse events that are estimated in the ACM test set (using only Spanish clinics) and those actually observed in the postmarketing cohort.

CONCLUSIONS

The risks of an ACM-MD can be accurately and robustly estimated, thus enhancing patients' safety. The proposed methodology is applicable to other clinical decisions based on predictive models since our proposal does not depend on the particular predictive model.

Enhanced prediction of hemoglobin concentration in a very large cohort of hemodialysis patients by means of deep recurrent neural networks

Erythropoiesis Stimulating Agents (ESAs) have become a standard anemia management tool for End Stage Renal Disease (ESRD) patients. However, dose optimization constitutes an extremely challenging task due to huge inter and intra-patient variability in the responses to ESA administration. Current data-based approaches to anemia control focus on learning accurate hemoglobin prediction models, which can be later utilized for testing competing treatment choices and choosing the optimal one. These methods, despite being proven effective in practice, present several shortcomings which this paper intends to tackle. Namely, they are limited to a small cohort of patients and, even then, they fail to provide suggestions when some strict requirements are not met (such as having a three month history prior to the prediction). Here, recurrent neural networks (RNNs) are used to model whole patient histories, providing predictions at every time step since the very first day. Furthermore, an unprecedented amount of data (∼110,000 patients from many different medical centers in twelve countries, without exclusion criteria) was used to train it, thus allowing it to generalize for every single patient. The resulting model outperforms state-of-the-art Hemoglobin prediction, providing excellent results even when tested on a prospective dataset. Simultaneously, it allows to bring the benefits of algorithmic anemia control to a very large group of patients.

Retrieving Quantum Information with Active Learning

Active learning is a machine learning method aiming at optimal design for model training. At variance with supervised learning, which labels all samples, active learning provides an improved model by labeling samples with maximal uncertainty according to the estimation model. Here, we propose the use of active learning for efficient quantum information retrieval, which is a crucial task in the design of quantum experiments. Meanwhile, when dealing with large data output, we employ active learning for the sake of classification with minimal cost in fidelity loss. Indeed, labeling only 5% samples, we achieve almost 90% rate estimation. The introduction of active learning methods in the data analysis of quantum experiments will enhance applications of quantum technologies.

Supervised Quantum Learning without Measurements

We propose a quantum machine learning algorithm for efficiently solving a class of problems encoded in quantum controlled unitary operations. The central physical mechanism of the protocol is the iteration of a quantum time-delayed equation that introduces feedback in the dynamics and eliminates the necessity of intermediate measurements. The performance of the quantum algorithm is analyzed by comparing the results obtained in numerical simulations with the outcome of classical machine learning methods for the same problem. The use of time-delayed equations enhances the toolbox of the field of quantum machine learning, which may enable unprecedented applications in quantum technologies.

A finite element-based machine learning approach for modeling the mechanical behavior of the breast tissues under compression in real-time

This work presents a data-driven method to simulate, in real-time, the biomechanical behavior of the breast tissues in some image-guided interventions such as biopsies or radiotherapy dose delivery as well as to speed up multimodal registration algorithms. Ten real breasts were used for this work. Their deformation due to the displacement of two compression plates was simulated off-line using the finite element (FE) method. Three machine learning models were trained with the data from those simulations. Then, they were used to predict in real-time the deformation of the breast tissues during the compression. The models were a decision tree and two tree-based ensemble methods (extremely randomized trees and random forest). Two different experimental setups were designed to validate and study the performance of these models under different conditions. The mean 3D Euclidean distance between nodes predicted by the models and those extracted from the FE simulations was calculated to assess the performance of the models in the validation set. The experiments proved that extremely randomized trees performed better than the other two models. The mean error committed by the three models in the prediction of the nodal displacements was under 2 mm, a threshold usually set for clinical applications. The time needed for breast compression prediction is sufficiently short to allow its use in real-time (<0.2 s).

Prediction of the hemoglobin level in hemodialysis patients using machine learning techniques

Patients who suffer from chronic renal failure (CRF) tend to suffer from an associated anemia as well. Therefore, it is essential to know the hemoglobin (Hb) levels in these patients. The aim of this paper is to predict the hemoglobin (Hb) value using a database of European hemodialysis patients provided by Fresenius Medical Care (FMC) for improving the treatment of this kind of patients. For the prediction of Hb, both analytical measurements and medication dosage of patients suffering from chronic renal failure (CRF) are used. Two kinds of models were trained, global and local models. In the case of local models, clustering techniques based on hierarchical approaches and the adaptive resonance theory (ART) were used as a first step, and then, a different predictor was used for each obtained cluster. Different global models have been applied to the dataset such as Linear Models, Artificial Neural Networks (ANNs), Support Vector Machines (SVM) and Regression Trees among others. Also a relevance analysis has been carried out for each predictor model, thus finding those features that are most relevant for the given prediction.

A new approach based on Machine Learning for predicting corneal curvature (K1) and astigmatism in patients with keratoconus after intracorneal ring implantation

Keratoconus (KC) is the most common type of corneal ectasia. A corneal transplantation was the treatment of choice until the last decade. However, intra-corneal ring implantation has become more and more common, and it is commonly used to treat KC thus avoiding a corneal transplantation. This work proposes a new approach based on Machine Learning to predict the vision gain of KC patients after ring implantation. That vision gain is assessed by means of the corneal curvature and the astigmatism. Different models were proposed; the best results were achieved by an artificial neural network based on the Multilayer Perceptron. The error provided by the best model was 0.97D of corneal curvature and 0.93D of astigmatism.

Optimization of anemia treatment in hemodialysis patients via reinforcement learning

OBJECTIVE

Anemia is a frequent comorbidity in hemodialysis patients that can be successfully treated by administering erythropoiesis-stimulating agents (ESAs). ESAs dosing is currently based on clinical protocols that often do not account for the high inter- and intra-individual variability in the patient's response. As a result, the hemoglobin level of some patients oscillates around the target range, which is associated with multiple risks and side-effects. This work proposes a methodology based on reinforcement learning (RL) to optimize ESA therapy.

METHODS

RL is a data-driven approach for solving sequential decision-making problems that are formulated as Markov decision processes (MDPs). Computing optimal drug administration strategies for chronic diseases is a sequential decision-making problem in which the goal is to find the best sequence of drug doses. MDPs are particularly suitable for modeling these problems due to their ability to capture the uncertainty associated with the outcome of the treatment and the stochastic nature of the underlying process. The RL algorithm employed in the proposed methodology is fitted Q iteration, which stands out for its ability to make an efficient use of data.

RESULTS

The experiments reported here are based on a computational model that describes the effect of ESAs on the hemoglobin level. The performance of the proposed method is evaluated and compared with the well-known Q-learning algorithm and with a standard protocol. Simulation results show that the performance of Q-learning is substantially lower than FQI and the protocol. When comparing FQI and the protocol, FQI achieves an increment of 27.6% in the proportion of patients that are within the targeted range of hemoglobin during the period of treatment. In addition, the quantity of drug needed is reduced by 5.13%, which indicates a more efficient use of ESAs.

CONCLUSION

Although prospective validation is required, promising results demonstrate the potential of RL to become an alternative to current protocols.

Estimation of the elastic parameters of human liver biomechanical models by means of medical images and evolutionary computation

This paper presents a method to computationally estimate the elastic parameters of two biomechanical models proposed for the human liver. The method is aimed at avoiding the invasive measurement of its mechanical response. The chosen models are a second order Mooney-Rivlin model and an Ogden model. A novel error function, the geometric similarity function (GSF), is formulated using similarity coefficients widely applied in the field of medical imaging (Jaccard coefficient and Hausdorff coefficient). This function is used to compare two 3D images. One of them corresponds to a reference deformation carried out over a finite element (FE) mesh of a human liver from a computer tomography image, whilst the other one corresponds to the FE simulation of that deformation in which variations in the values of the model parameters are introduced. Several search strategies, based on GSF as cost function, are developed to accurately find the elastics parameters of the models, namely: two evolutionary algorithms (scatter search and genetic algorithm) and an iterative local optimization. The results show that GSF is a very appropriate function to estimate the elastic parameters of the biomechanical models since the mean of the relative mean absolute errors committed by the three algorithms is lower than 4%.

Profiled support vector machines for antisense oligonucleotide efficacy prediction

BACKGROUND

This paper presents the use of Support Vector Machines (SVMs) for prediction and analysis of antisense oligonucleotide (AO) efficacy. The collected database comprises 315 AO molecules including 68 features each, inducing a problem well-suited to SVMs. The task of feature selection is crucial given the presence of noisy or redundant features, and the well-known problem of the curse of dimensionality. We propose a two-stage strategy to develop an optimal model: (1) feature selection using correlation analysis, mutual information, and SVM-based recursive feature elimination (SVM-RFE), and (2) AO prediction using standard and profiled SVM formulations. A profiled SVM gives different weights to different parts of the training data to focus the training on the most important regions.

RESULTS

In the first stage, the SVM-RFE technique was most efficient and robust in the presence of low number of samples and high input space dimension. This method yielded an optimal subset of 14 representative features, which were all related to energy and sequence motifs. The second stage evaluated the performance of the predictors (overall correlation coefficient between observed and predicted efficacy, r; mean error, ME; and root-mean-square-error, RMSE) using 8-fold and minus-one-RNA cross-validation methods. The profiled SVM produced the best results (r = 0.44, ME = 0.022, and RMSE= 0.278) and predicted high (>75% inhibition of gene expression) and low efficacy (<25%) AOs with a success rate of 83.3% and 82.9%, respectively, which is better than by previous approaches. A web server for AO prediction is available online at http://aosvm.cgb.ki.se/.

CONCLUSIONS

The SVM approach is well suited to the AO prediction problem, and yields a prediction accuracy superior to previous methods. The profiled SVM was found to perform better than the standard SVM, suggesting that it could lead to improvements in other prediction problems as well.