Automated recognition of functioning, activity and participation in COVID-19 from electronic patient records by natural language processing: a proof- of- concept.

PURPOSE: To address the feasibility, reliability and internal validity of natural language processing (NLP) for automated functional assessment of hospitalised COVID-19 patients in key International Classification of Functioning, Disability and Health (ICF) categories and levels from unstructured text in electronic health records (EHR) from a large teaching hospital. MATERIALS AND METHODS: Eight human annotators assigned four ICF categories to relevant sentences: Emotional functions, Exercise tolerance, Walking and Moving, Work and Employment and their ICF levels (Functional Ambulation Categories for Walking and Moving, metabolic equivalents for Exercise tolerance). A linguistic neural network-based model was trained on 80% of the annotated sentences; inter-annotator agreement (IAA, Cohen's kappa), a weighted score of precision and recall (F1) and RMSE for level detection were assessed for the remaining 20%. RESULTS: In total 4112 sentences of non-COVID-19 and 1061 of COVID-19 patients were annotated. Average IAA was 0.81; F1 scores were 0.7 for Walking and Moving and Emotional functions; RMSE for Walking and Moving (5- level scale) was 1.17 for COVID-19 patients. CONCLUSION: Using a limited amount of annotated EHR sentences, a proof-of-concept was obtained for automated functional assessment of COVID-19 patients in ICF categories and levels. This allows for instantaneous assessment of the functional consequences of new diseases like COVID-19 for large numbers of patients.Key messagesHospitalised Covid-19 survivors may persistently suffer from low physical and mental functioning and a reduction in overall quality of life requiring appropriate and personalised rehabilitation strategies.For this, assessment of functioning within multiple domains and categories of the International Classification of Function is required, which is cumbersome using structured data.We show a proof-of-concept using Natural Language Processing techniques to automatically derive the aforementioned information from free-text notes within the Electronic Health Record of a large academic teaching hospital.

Action of the polyamine beta-philanthotoxin on neuromuscular transmission in insects.

The beta-fraction of the venom of the solitary wasp Philanthus triangulum inhibited glutamate-induced potentials in muscle fibres of the locust in a dose-dependent manner (50% block at 2 micrograms/ml). Single channel behaviour and miniature excitatory postsynaptic currents were not affected. The blocking effect disappeared after treatment of the muscle with the lectin concanavalin A which prevents desensitization of glutamate receptors. Block of potentials induced by iontophoretic application of glutamate was strongly dependent on the frequency of stimulation: no block occurred at stimulus intervals of 30 sec or more. The nature of this inhibition is discussed and proposed to be a delay of recovery from desensitization.

Selectivity of the uptake of glutamate and GABA in two morphologically distinct insect neuromuscular synapses.

The common inhibitor (CI) and slow excitor tibiae (SETi) innervated slow muscles 135cd of the locust Schistocerca gregaria were incubated under high-affinity uptake conditions either in [3H]GABA or in [3H]glutamate. [3H]GABA is accumulated in the glia of the nerve endings of the CI as well as the SETi; however, it is accumulated only in the terminal axons of the CI, not in the terminal axons of the SETi. The grain densities above the glia and above the CI terminal axons are approximately 2 grains/micron2. After incubation in [3H]glutamate the grain densities above the CI terminal axons and the SETi terminal axons are approximately 4 grains/micron2; the grain densities above the glia of both types of nerve endings are approximately 17 grains/micron2. The relatively high labeling (3 grains/micron2) of the muscles after incubation in the presence of glutamate is ascribed to the high metabolic requirements of slow muscles. The conclusion is drawn that a high-affinity uptake system for GABA is present in the CI terminal axons and in the glia of both the CI and SETi nerve endings. However, while the glutamate uptake in the CI and SETi nerve endings of the slow 135cd is comparable to the high-affinity uptake of glutamate in the fast excitor tibiae (FETi) nerve endings of the fast retractor unguis muscle, a high-affinity uptake of glutamate was only demonstrated in the glia of both types of nerve endings. A high-affinity uptake in the terminal axons of the CI and SETi may be masked by an extensively low-affinity uptake of glutamate by the muscles.

Block of locust muscle glutamate receptors by delta-philanthotoxin occurs after receptor activations.

One component (delta-philanthotoxin (delta-PTX) of the venom from the wasp Philanthus triangulum blocks transmission postsynaptically at excitatory synapses on locust muscle. delta-PTX depresses both the iontophoretic glutamate potential and the excitatory junctional current (e.j.c.) in a glutamate receptor activation-dependent manner. The rate of recovery from the effects of the toxin is reduced following either prolonged application of L-glutamate or repetitive iontophoretic application of this amino acid or high frequency neural stimulation of the muscle in the presence of delta-PTX. The decay phase of the e.j.c. is shortened by delta-PTX. The effects of delta-PTX on the e.j.c. are not voltage dependent. The open-close kinetics of glutamate channels in extrajunctional muscle membrane are modified by delta-PTX as shown by patch clamp analysis. The mean life time of the glutamate channel is reduced, whilst the mean interval between single opening events is increased with the events often occurring in bursts. These data are consistent with glutamate channel blocking by this toxin. It is proposed that the toxin blocks open channels gated by both junctional and extrajunctional glutamate receptors on locust muscle. It is further proposed that delta-PTX enters a compartment of the muscle through the glutamate open channels and that it can also block the open channels from this site.

Philanthotoxin inhibits Ca2+ currents in rat hippocampal CA1 neurons.

The wasp venom philanthotoxin-4.3.3 (PhTX-4.3.3) is an antagonist of glutamate transmission in the insect as well as in the mammalian brain. It was recently shown that PhTX-4.3.3 inhibits the N-methyl-D-aspartate (NMDA) transmission in rat hippocampus. In this study we show that dideaza-philanthotoxin-12 (dideaza-PhTX-12), an analogue of PhTX-4.3.3, is a potent antagonist of voltage-dependent Ca2+ currents in rat hippocampal CA1 neurons. At a concentration of 10 microM it reduces the Ca2+ current to 40%. Two voltage-dependent potassium currents, the A current and the delayed rectifier, were hardly affected by dideaza-PhTX-12, indicating selectivity of the drug for Ca2+ currents. As a consequence the philanthotoxins will inhibit the calcium influx via voltage dependent as well as NMDA mediaded calcium channels and thus reduce excitability in the hippocampus.

The benzodiazepine midazolam preferentially blocks inactivated Na channels in skeletal muscle fibre.

The effects of the benzodiazepine midazolam were studied on frog skeletal muscle fibres held under current- or voltage-clamp conditions. Midazolam induced a concentration-dependent (10(-5) mol/l to 10(-3) mol/l) block of the action potential and of the underlying Na current. Block of the Na current occurred without any changes in its voltage dependence or in its activation and inactivation kinetics. An apparent dissociation constant of 223 mumol/l was determined for midazolam from the rested Na channels of well polarized fibres. The blocking effect of a threshold concentration (10(-5) mol/l) could be greatly enhanced (up to the complete suppression of the current) by predepolarizations, positive holding potentials or high stimulation frequencies. This apparent voltage- and frequency-dependent block (no use dependence, i.e., no activation block) could be ascribed to a blockade of inactivated Na channels. From the apparent shift towards negative potentials of the steady-state inactivation curve, a dissociation constant of 6.0 mumol/l was calculated for midazolam from the inactivated Na channels, according to the modulated-receptor model. These results show that midazolam preferentially blocks inactivated rather than rested Na channels, and suggest that this mechanism of action might contribute to the well-known myorelaxant effect of the benzodiazepines.

Neurotoxic kinins from wasp and ant venoms.

Kinins are polypeptides of 9-18 amino-acid residues containing a bradykinin-like sequence, in some cases as part of a molecule. The bradykinin-like sequence is either bradykinin, Hyp3-bradykinin or Thr6-bradykinin. Kinins are neurotoxic components of wasp and ant venoms, causing in the insect CNS a presynaptic block of the cholinergic transmission by means of an irreversible depletion, probably caused by a non-competitive inhibition of choline uptake.

Two kinins isolated from an extract of the venom reservoirs of the solitary wasp Megascolia flavifrons.

From an extract of the venom reservoirs of the wasp Megascolia flavifrons two kinins have been isolated. The sequences of amino acids are: Arg-Pro-Pro-Gly-Phe-Thr-Pro-Phe-Arg (Thr6-bradykinin) and Arg-Pro-Pro-Gly-Phe-Thr-Pro-Phe-Arg-Lys-Ala (Thr6-bradykinin-Lys-Ala). The bradykinin-like effects of the venom on a number of vertebrate smooth muscle preparations can be explained by the actions of these kinins.

Comparison of the immunogenicity of wasp venom peptides with or without carbohydrate moieties.

Three synthetic vespulakinin analogues either with or without carbohydrate moieties and mastoparan B isolated from Vespa basalis venom were investigated for their immunogenic activity and solution conformation. Mice immunized with these wasp venom peptides, with the exception of (Gal alpha)Thr3, (Gal alpha)Thr4-vespulakinin 1, showed positive antibody responses. However, the response elicited by mastoparan B was much higher than those induced by vespulakinin analogues. The class of antibody induced by these peptides was identified as an IgG1 isotype with kappa-light chain, suggesting stimulation of a T-cell-dependent immune response by these peptides. According to the circular dichroism spectra of these peptides, the structures of the vespulakinin analogues in solution were largely unordered, while mastoparan B exhibited a conformation rich in alpha-helices. The presence of carbohydrate moieties and the rather random structure in vespulakinins may interfere with T-cell recognition of the peptides, leading to lower immune responses.

B2-kininergic action of linear and cyclic tryptophan6- and tyrosine6-kallidin.

This study was undertaken to determine the importance of the central aromatic moiety in the kallidin and cyclokallidin molecules, using the relaxation of the isolated duodenum of the rat. Replacement in kallidin of the central phenylalanine by tryptophan increased the potency from an EC50 of 3 x 10(-10)M to 2 x 10(-12)M. Replacement by tyrosine decreased the potency to an EC50 of 8 x 10(-8)M. In cyclo-kallidin (EC50: 10(-8)M) the potencies were decreased: cyclo-Trp6-kallidin showed an EC50 of 10(-6)M and cyclo-Tyr6-kallidin of 3 x 10(-7)M. The relaxation of the rat duodenum by linear and cyclic kinins was potentiated by the bradykinin potentiating peptide BPP5a and antagonized by the B2 antagonist HOE-140. At a concentration of 10(-9)M, HOE-140 significantly decreased the potencies of bradykinin and cyclo-kallidin, but not of the B1 agonist desArg9-bradykinin.