Development and validation of machine learning algorithms based on electrocardiograms for cardiovascular diagnoses at the population level.

Artificial intelligence-enabled electrocardiogram (ECG) algorithms are gaining prominence for the early detection of cardiovascular (CV) conditions, including those not traditionally associated with conventional ECG measures or expert interpretation. This study develops and validates such models for simultaneous prediction of 15 different common CV diagnoses at the population level. We conducted a retrospective study that included 1,605,268 ECGs of 244,077 adult patients presenting to 84 emergency departments or hospitals, who underwent at least one 12-lead ECG from February 2007 to April 2020 in Alberta, Canada, and considered 15 CV diagnoses, as identified by International Classification of Diseases, 10th revision (ICD-10) codes: atrial fibrillation (AF), supraventricular tachycardia (SVT), ventricular tachycardia (VT), cardiac arrest (CA), atrioventricular block (AVB), unstable angina (UA), ST-elevation myocardial infarction (STEMI), non-STEMI (NSTEMI), pulmonary embolism (PE), hypertrophic cardiomyopathy (HCM), aortic stenosis (AS), mitral valve prolapse (MVP), mitral valve stenosis (MS), pulmonary hypertension (PHTN), and heart failure (HF). We employed ResNet-based deep learning (DL) using ECG tracings and extreme gradient boosting (XGB) using ECG measurements. When evaluated on the first ECGs per episode of 97,631 holdout patients, the DL models had an area under the receiver operating characteristic curve (AUROC) of <80% for 3 CV conditions (PTE, SVT, UA), 80-90% for 8 CV conditions (CA, NSTEMI, VT, MVP, PHTN, AS, AF, HF) and an AUROC > 90% for 4 diagnoses (AVB, HCM, MS, STEMI). DL models outperformed XGB models with about 5% higher AUROC on average. Overall, ECG-based prediction models demonstrated good-to-excellent prediction performance in diagnosing common CV conditions.

Towards artificial intelligence-based learning health system for population-level mortality prediction using electrocardiograms.

The feasibility and value of linking electrocardiogram (ECG) data to longitudinal population-level administrative health data to facilitate the development of a learning healthcare system has not been fully explored. We developed ECG-based machine learning models to predict risk of mortality among patients presenting to an emergency department or hospital for any reason. Using the 12-lead ECG traces and measurements from 1,605,268 ECGs from 748,773 healthcare episodes of 244,077 patients (2007-2020) in Alberta, Canada, we developed and validated ResNet-based Deep Learning (DL) and gradient boosting-based XGBoost (XGB) models to predict 30-day, 1-year, and 5-year mortality. The models for 30-day, 1-year, and 5-year mortality were trained on 146,173, 141,072, and 111,020 patients and evaluated on 97,144, 89,379, and 55,650 patients, respectively. In the evaluation cohort, 7.6%, 17.3%, and 32.9% patients died by 30-days, 1-year, and 5-years, respectively. ResNet models based on ECG traces alone had good-to-excellent performance with area under receiver operating characteristic curve (AUROC) of 0.843 (95% CI: 0.838-0.848), 0.812 (0.808-0.816), and 0.798 (0.792-0.803) for 30-day, 1-year and 5-year prediction, respectively; and were superior to XGB models based on ECG measurements with AUROC of 0.782 (0.776-0.789), 0.784 (0.780-0.788), and 0.746 (0.740-0.751). This study demonstrates the validity of ECG-based DL mortality prediction models at the population-level that can be leveraged for prognostication at point of care.

Nature of bilayer lipids affects membranes deformation and pore resealing during nanoparticle penetration.

Interactions of nanoparticles (NPs) with lipid membranes have enormous biological implications especially for gene delivery applications. In this work, using all-atom steered- and molecular dynamics simulations, we investigated deformation of lipid membranes and pore closure during a NP penetration process. Three membrane bilayer models built from 2-oleoyl-1-palmitoyl-sn-glycero-3-phosphocholine (POPC), dipalmitoylphosphatidylcholine (DPPC) and dilauroylphosphatidylcholine (DLPC), and a NP formed by 2 short interfering RNA (siRNA) and 6 polyethylenimine (PEI) molecules were used. Our results showed that different membrane lipids could lead to differences in pore formation (symmetric vs. asymmetric), and could undergo different levels of pore-mediated flip-flops during the closure. DLPC showed the largest number of flip-flops among the three lipid membranes. In addition, introduction of hydrophobic linoleic acid (LA) substitution onto the PEIs was found to facilitate pore formation, since the long LA tails could insert themselves into the hydrophobic region of the membrane where the lipid tails were less aligned. Compared with DPPC, POPC and DLPC membranes had less alignment of lipid tails in the bilayer, which promoted the insertion of LA tails and hence NP entry into the cell. Our observations provide valuable insight into the membrane deformations and pore dynamics during NP penetration and will be important for the design of NP carriers for effective gene delivery.

Membrane lipids destabilize short interfering ribonucleic acid (siRNA)/polyethylenimine nanoparticles.

Cell entry of polymeric nanoparticles (NPs) bearing polynucleotides is an important stage for successful gene delivery. In this work, we addressed the influence of cell membrane lipids on the integrity and configurational changes of NPs composed of short interfering ribonucleic acid (siRNA) and polyethylenimine. We focused on NPs derived from two different PEIs, unmodified low molecular weight PEI and linoleic acid (LA)-substituted PEI, and their interactions with two membrane lipids (zwitterionic 2-oleoyl-1-palmitoyl-sn-glycero-3-phosphocholine (POPC) and anionic 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-l-serine (POPS)). Our experiments showed that POPS liposomes interacted strongly with both types of NPs, which caused partial dissociation of the NPs. POPC liposomes, however, did not induce any dissociation. Consistent with the experiments, steered molecular dynamics simulations showed a stronger interaction between the NPs and the POPS membrane than between the NPs and the POPC membrane. Lipid substitution on the PEIs enhanced the stability of the NPs during membrane crossing; lipid association between PEIs of the LA-bearing NPs as well as parallel orientation of the siRNAs provided protection against their dissociation (unlike NPs from native PEI). Our observations provide valuable insight into the integrity and structural changes of PEI/siRNA NPs during membrane crossing which will help in the design of more effective carriers for nucleic acid delivery.

Steered molecular dynamics simulations reveal a self-protecting configuration of nanoparticles during membrane penetration.

Cell entry of polynucleotide-based therapeutic agents can be facilitated by nanoparticle (NP) mediated delivery. In this work, using steered molecular dynamics simulations, we simulated the membrane penetration process of a NP formed by 2 short interfering RNA (siRNA) and 6 polyethylenimine (PEI) molecules. To the best of our knowledge, this is the first set of simulations that explore the direct penetration of an siRNA/PEI NP through a membrane at an all-atom scale. Three types of PEI molecules were used for NP formation: a native PEI, a PEI modified with caprylic acids and a PEI modified with linoleic acids. We found that hydrogen bond formation between the PEIs and the membrane did not lead to instability of the siRNA/PEI NPs during the internalization process. Instead, our results suggested adoption of a "self-protecting" configuration by the siRNA/PEI NP during membrane penetration, where the siRNA/PEI NP becomes more compact and siRNAs become aligned, leading to more stable configurations while detaching from the membrane. The siRNA/PEI NP modified with linoleic acid showed the smallest structural change due to its strong intra-particle lipid associations and the resulting rigidity, while NP modified with caprylic acid showed the largest structural changes. Our observations provide unique insight into the structural changes of siRNA/PEI NPs when crossing the cell membrane, which can be important for the design of new NP carriers for nucleic acid delivery.

Carbon nanotube-encapsulated drug penetration through the cell membrane: an investigation based on steered molecular dynamics simulation.

Understanding the penetration mechanisms of carbon nanotube (CNTs)-encapsulated drugs through the phospholipid bilayer cell membrane is an important issue for the development of intracellular drug delivery systems. In the present work, steered molecular dynamics (SMD) simulation was used to explore the possibility of penetration of a polar drug, paclitaxel (PTX), encapsulated inside the CNT, through a dipalmitoylphosphatidylcholine bilayer membrane. The interactions between PTX and CNT and between PTX and the confined water molecules inside the CNT had a significant effect on the penetration process of PTX. The results reveal that the presence of a PTX molecule increases the magnitude of the pulling force. The effect of pulling velocity on the penetration mechanism was also investigated by a series of SMD simulations, and it is shown that the pulling velocity had a significant effect on pulling force and the interaction between lipid bilayer and drug molecule.