A Systematic Review of Time Series Classification Techniques Used in Biomedical Applications.

Background: Digital clinical measures collected via various digital sensing technologies such as smartphones, smartwatches, wearables, and ingestible and implantable sensors are increasingly used by individuals and clinicians to capture the health outcomes or behavioral and physiological characteristics of individuals. Time series classification (TSC) is very commonly used for modeling digital clinical measures. While deep learning models for TSC are very common and powerful, there exist some fundamental challenges. This review presents the non-deep learning models that are commonly used for time series classification in biomedical applications that can achieve high performance. Objective: We performed a systematic review to characterize the techniques that are used in time series classification of digital clinical measures throughout all the stages of data processing and model building. Methods: We conducted a literature search on PubMed, as well as the Institute of Electrical and Electronics Engineers (IEEE), Web of Science, and SCOPUS databases using a range of search terms to retrieve peer-reviewed articles that report on the academic research about digital clinical measures from a five-year period between June 2016 and June 2021. We identified and categorized the research studies based on the types of classification algorithms and sensor input types. Results: We found 452 papers in total from four different databases: PubMed, IEEE, Web of Science Database, and SCOPUS. After removing duplicates and irrelevant papers, 135 articles remained for detailed review and data extraction. Among these, engineered features using time series methods that were subsequently fed into widely used machine learning classifiers were the most commonly used technique, and also most frequently achieved the best performance metrics (77 out of 135 articles). Statistical modeling (24 out of 135 articles) algorithms were the second most common and also the second-best classification technique. Conclusions: In this review paper, summaries of the time series classification models and interpretation methods for biomedical applications are summarized and categorized. While high time series classification performance has been achieved in digital clinical, physiological, or biomedical measures, no standard benchmark datasets, modeling methods, or reporting methodology exist. There is no single widely used method for time series model development or feature interpretation, however many different methods have proven successful.

Rapid monophasic HBsAg decline during nucleic-acid polymer-based therapy predicts functional cure.

BACKGROUND AND AIMS: Analyzing the interplay among serum HBV DNA, HBsAg, anti-HBs, and alanine aminotransferase (ALT) during nucleic-acid polymer (NAP)-based therapy for chronic hepatitis B provides a unique opportunity to identify kinetic patterns associated with functional cure. METHODS: All participants with HBeAg-negative chronic HBV infection in the REP 401 study (NCT02565719) first received 24 weeks of tenofovir-disoproxil-fumarate (TDF) monotherapy. The early triple therapy group (n = 20) next received 48 weeks of TDF+pegylated interferon-α2a (pegIFN)+NAPs. In contrast, the delayed triple therapy group (n = 20) next received 24 weeks of TDF+pegIFN before 48 weeks of triple therapy. Three participants discontinued treatment and were excluded. Functional cure (HBsAg and HBV DNA not detectable with normal ALT) was assessed at 48 weeks post-treatment. Different kinetic phases were defined by at least a 2-fold change in slope. A single-phase decline was categorized as monophasic, and 2-phase declines were categorized as biphasic. RESULTS: Fourteen (35%) participants achieved a functional cure. HBV DNA remained below or near undetectable for all participants by the end of TDF monotherapy and during subsequent combination therapies. Three HBsAg kinetic patterns were found in both the early and delayed groups, nonresponders (n = 4 and n = 4), monophasic (n = 11 and n = 11), and biphasic (n = 4 and n = 3), respectively. All participants who achieved a functional cure had a monophasic HBsAg kinetic pattern during triple therapy. Among participants with a monophasic HBsAg decline, those who had a functional cure had a shorter median time to HBsAg loss of 21 (interquartile range=11) weeks compared with those who did not achieve functional cure [median: 27 (7) weeks] (p = 0.012). CONCLUSIONS: Functional cure was associated with a rapid monophasic HBsAg decline during NAP-based therapy. A nonmonophasic HBsAg kinetic pattern had a 100% negative predictive value (NPV) for a functional cure.

Hepatitis delta virus RNA decline post inoculation in human NTCP transgenic mice is biphasic.

Background and Aims: Chronic infection with hepatitis B and hepatitis delta viruses (HDV) is considered the most serious form of viral hepatitis due to more severe manifestations of and accelerated progression to liver fibrosis, cirrhosis, and hepatocellular carcinoma. There is no FDA-approved treatment for HDV and current interferon-alpha treatment is suboptimal. We characterized early HDV kinetics post inoculation and incorporated mathematical modeling to provide insights into host-HDV dynamics. Methods: We analyzed HDV RNA serum viremia in 192 immunocompetent (C57BL/6) and immunodeficient (NRG) mice that did or did not transgenically express the HDV receptor - human sodium taurocholate co-transporting peptide (hNTCP). Results: Kinetic analysis indicates an unanticipated biphasic decline consisting of a sharp first-phase and slower second-phase decline regardless of immunocompetence. HDV decline after re-inoculation again followed a biphasic decline; however, a steeper second-phase HDV decline was observed in NRG-hNTCP mice compared to NRG mice. HDV-entry inhibitor bulevirtide administration and HDV re-inoculation indicated that viral entry and receptor saturation are not major contributors to clearance, respectively. The biphasic kinetics can be mathematically modeled by assuming the existence of a non-specific binding compartment with a constant on/off-rate and the steeper second-phase decline by a loss of bound virus that cannot be returned as free virus to circulation. The model predicts that free HDV is cleared with a half-life of 18 minutes (standard error, SE: 2.4), binds to non-specific cells with a rate of 0.06 hour -1 (SE: 0.03), and returns as free virus with a rate of 0.23 hour -1 (SE: 0.03). Conclusions: Understanding early HDV-host kinetics will inform pre-clinical therapeutic kinetic studies on how the efficacy of anti-HDV therapeutics can be affected by early kinetics of viral decline. LAY SUMMARY: The persistence phase of HDV infection has been studied in some animal models, however, the early kinetics of HDV in vivo is incompletely understood. In this study, we characterize an unexpectedly HDV biphasic decline post inoculation in immunocompetent and immunodeficient mouse models and use mathematical modeling to provide insights into HDV-host dynamics. Understanding the kinetics of viral clearance in the blood can aid pre-clinical development and testing models for anti-HDV therapeutics.

Hepatitis delta virus RNA decline post-inoculation in human NTCP transgenic mice is biphasic.

Chronic infection with hepatitis B and delta viruses (HDV) is the most serious form of viral hepatitis due to more severe manifestations of an accelerated progression to liver fibrosis, cirrhosis, and hepatocellular carcinoma. We characterized early HDV kinetics post-inoculation and incorporated mathematical modeling to provide insights into host-HDV dynamics. We analyzed HDV RNA serum viremia in 192 immunocompetent (C57BL/6) and immunodeficient (NRG) mice that did or did not transgenically express the HDV receptor-human sodium taurocholate co-transporting polypeptide (hNTCP). Kinetic analysis indicates an unanticipated biphasic decline consisting of a sharp first-phase and slower second-phase decline regardless of immunocompetence. HDV decline after re-inoculation again followed a biphasic decline; however, a steeper second-phase HDV decline was observed in NRG-hNTCP mice compared to NRG mice. HDV-entry inhibitor bulevirtide administration and HDV re-inoculation indicated that viral entry and receptor saturation are not major contributors to clearance, respectively. The biphasic kinetics can be mathematically modeled by assuming the existence of a non-specific-binding compartment with a constant on/off-rate and the steeper second-phase decline by a loss of bound virus that cannot be returned as free virus to circulation. The model predicts that free HDV is cleared with a half-life of 35 minutes (standard error, SE: 6.3), binds to non-specific cells with a rate of 0.05 per hour (SE: 0.01), and returns as free virus with a rate of 0.11 per hour (SE: 0.02). Characterizing early HDV-host kinetics elucidates how quickly HDV is either cleared or bound depending on the immunological background and hNTCP presence. IMPORTANCE The persistence phase of HDV infection has been studied in some animal models; however, the early kinetics of HDV in vivo is incompletely understood. In this study, we characterize an unexpectedly HDV biphasic decline post-inoculation in immunocompetent and immunodeficient mouse models and use mathematical modeling to provide insights into HDV-host dynamics.

Modelling hepatitis D virus RNA and HBsAg dynamics during nucleic acid polymer monotherapy suggest rapid turnover of HBsAg.

Hepatitis D virus (HDV) requires hepatitis B surface antigen (HBsAg) for its assembly and release. Current HBV treatments are only marginally effective against HDV because they fail to inhibit HBsAg production/secretion. However, monotherapy with the nucleic acid polymer REP 2139-Ca is accompanied by rapid declines in both HBsAg and HDV RNA. We used mathematical modeling to estimate HDV-HBsAg-host parameters and to elucidate the mode of action and efficacy of REP 2139-Ca against HDV in 12 treatment-naive HBV/HDV co-infected patients. The model accurately reproduced the observed decline of HBsAg and HDV, which was simultaneous. Median serum HBsAg half-life (t1/2) was estimated as 1.3 [0.9-1.8] days corresponding to a pretreatment production and clearance of ~108 [107.7-108.3] IU/day. The HDV-infected cell loss was estimated to be 0.052 [0.035-0.074] days-1 corresponding to an infected cell t1/2 = 13.3 days. The efficacy of blocking HBsAg and HDV production were 98.2 [94.5-99.9]% and 99.7 [96.0-99.8]%, respectively. In conclusion, both HBsAg production and HDV replication are effectively inhibited by REP 2139-Ca. Modeling HBsAg kinetics during REP 2139-Ca monotherapy indicates a short HBsAg half-life (1.3 days) suggesting a rapid turnover of HBsAg in HBV/HDV co-infection.