Timesynth: A Temporal Fidelity Framework for Health Signal Digital Twins. Haque, M. R., Elhabian, S., & Pettine, W. W. July, 2026. arXiv:2607.00431 [cs.LG]![Timesynth: A Temporal Fidelity Framework for Health Signal Digital Twins [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex Forecasting models for health-signal digital twins must preserve the oscillatory, frequency, phase, and state-transition dynamics of physiological signals, yet the pointwise metrics used to benchmark them cannot detect when these fundamental properties are lost. We show that this blind spot misranks models: across 11 architectures, models with comparable pointwise error diverge by up to 53° in phase accuracy, equivalent to roughly 123 ms for a 1.2 Hz cardiac rhythm and invisible to standard metrics. To enable development of models that escape such failures, we introduce TimeSynth, a controlled benchmarking framework with two reusable components: a physiologically grounded generator producing signals with analytically known ground-truth dynamics from parametric models fitted to real electroencephalography, electrocardiography and photoplethysmogram signals, along with diagnostics quantifying amplitude, frequency, phase, and state-transition fidelity. Linear and full-sequence attention models systematically lose frequency and phase information despite acceptable amplitude error, whereas architectures with localized temporal structure better preserve dynamical fidelity and adapt to observable state transitions; none, however, reliably preserves stochastic switching. Because the dominant determinant of fidelity is architectural, model choice becomes a principled, use-case-driven decision rather than a search for a single winner. TimeSynth thus supplies the controlled preclinical stress test missing before models are coupled to patient data, with a reusable generator and diagnostics for fidelity-aware development.
@misc{haque_timesynth_2026,
title = {Timesynth: {A} {Temporal} {Fidelity} {Framework} for {Health} {Signal} {Digital} {Twins}},
shorttitle = {Timesynth},
url = {http://arxiv.org/abs/2607.00431},
doi = {10.48550/arXiv.2607.00431},
abstract = {Forecasting models for health-signal digital twins must preserve the oscillatory, frequency, phase, and state-transition dynamics of physiological signals, yet the pointwise metrics used to benchmark them cannot detect when these fundamental properties are lost. We show that this blind spot misranks models: across 11 architectures, models with comparable pointwise error diverge by up to 53° in phase accuracy, equivalent to roughly 123 ms for a 1.2 Hz cardiac rhythm and invisible to standard metrics. To enable development of models that escape such failures, we introduce TimeSynth, a controlled benchmarking framework with two reusable components: a physiologically grounded generator producing signals with analytically known ground-truth dynamics from parametric models fitted to real electroencephalography, electrocardiography and photoplethysmogram signals, along with diagnostics quantifying amplitude, frequency, phase, and state-transition fidelity. Linear and full-sequence attention models systematically lose frequency and phase information despite acceptable amplitude error, whereas architectures with localized temporal structure better preserve dynamical fidelity and adapt to observable state transitions; none, however, reliably preserves stochastic switching. Because the dominant determinant of fidelity is architectural, model choice becomes a principled, use-case-driven decision rather than a search for a single winner. TimeSynth thus supplies the controlled preclinical stress test missing before models are coupled to patient data, with a reusable generator and diagnostics for fidelity-aware development.},
urldate = {2026-07-28},
publisher = {arXiv},
author = {Haque, Md Rakibul and Elhabian, Shireen and Pettine, Warren Woodrich},
month = jul,
year = {2026},
note = {arXiv:2607.00431 [cs.LG]},
keywords = {Computer Science - Machine Learning},
}
Downloads: 0
{"_id":"cL4tNE8eq4ZMqAkfn","bibbaseid":"haque-elhabian-pettine-timesynthatemporalfidelityframeworkforhealthsignaldigitaltwins-2026","author_short":["Haque, M. R.","Elhabian, S.","Pettine, W. W."],"bibdata":{"bibtype":"misc","type":"misc","title":"Timesynth: A Temporal Fidelity Framework for Health Signal Digital Twins","shorttitle":"Timesynth","url":"http://arxiv.org/abs/2607.00431","doi":"10.48550/arXiv.2607.00431","abstract":"Forecasting models for health-signal digital twins must preserve the oscillatory, frequency, phase, and state-transition dynamics of physiological signals, yet the pointwise metrics used to benchmark them cannot detect when these fundamental properties are lost. We show that this blind spot misranks models: across 11 architectures, models with comparable pointwise error diverge by up to 53° in phase accuracy, equivalent to roughly 123 ms for a 1.2 Hz cardiac rhythm and invisible to standard metrics. To enable development of models that escape such failures, we introduce TimeSynth, a controlled benchmarking framework with two reusable components: a physiologically grounded generator producing signals with analytically known ground-truth dynamics from parametric models fitted to real electroencephalography, electrocardiography and photoplethysmogram signals, along with diagnostics quantifying amplitude, frequency, phase, and state-transition fidelity. Linear and full-sequence attention models systematically lose frequency and phase information despite acceptable amplitude error, whereas architectures with localized temporal structure better preserve dynamical fidelity and adapt to observable state transitions; none, however, reliably preserves stochastic switching. Because the dominant determinant of fidelity is architectural, model choice becomes a principled, use-case-driven decision rather than a search for a single winner. TimeSynth thus supplies the controlled preclinical stress test missing before models are coupled to patient data, with a reusable generator and diagnostics for fidelity-aware development.","urldate":"2026-07-28","publisher":"arXiv","author":[{"propositions":[],"lastnames":["Haque"],"firstnames":["Md","Rakibul"],"suffixes":[]},{"propositions":[],"lastnames":["Elhabian"],"firstnames":["Shireen"],"suffixes":[]},{"propositions":[],"lastnames":["Pettine"],"firstnames":["Warren","Woodrich"],"suffixes":[]}],"month":"July","year":"2026","note":"arXiv:2607.00431 [cs.LG]","keywords":"Computer Science - Machine Learning","bibtex":"@misc{haque_timesynth_2026,\n\ttitle = {Timesynth: {A} {Temporal} {Fidelity} {Framework} for {Health} {Signal} {Digital} {Twins}},\n\tshorttitle = {Timesynth},\n\turl = {http://arxiv.org/abs/2607.00431},\n\tdoi = {10.48550/arXiv.2607.00431},\n\tabstract = {Forecasting models for health-signal digital twins must preserve the oscillatory, frequency, phase, and state-transition dynamics of physiological signals, yet the pointwise metrics used to benchmark them cannot detect when these fundamental properties are lost. We show that this blind spot misranks models: across 11 architectures, models with comparable pointwise error diverge by up to 53° in phase accuracy, equivalent to roughly 123 ms for a 1.2 Hz cardiac rhythm and invisible to standard metrics. To enable development of models that escape such failures, we introduce TimeSynth, a controlled benchmarking framework with two reusable components: a physiologically grounded generator producing signals with analytically known ground-truth dynamics from parametric models fitted to real electroencephalography, electrocardiography and photoplethysmogram signals, along with diagnostics quantifying amplitude, frequency, phase, and state-transition fidelity. Linear and full-sequence attention models systematically lose frequency and phase information despite acceptable amplitude error, whereas architectures with localized temporal structure better preserve dynamical fidelity and adapt to observable state transitions; none, however, reliably preserves stochastic switching. Because the dominant determinant of fidelity is architectural, model choice becomes a principled, use-case-driven decision rather than a search for a single winner. TimeSynth thus supplies the controlled preclinical stress test missing before models are coupled to patient data, with a reusable generator and diagnostics for fidelity-aware development.},\n\turldate = {2026-07-28},\n\tpublisher = {arXiv},\n\tauthor = {Haque, Md Rakibul and Elhabian, Shireen and Pettine, Warren Woodrich},\n\tmonth = jul,\n\tyear = {2026},\n\tnote = {arXiv:2607.00431 [cs.LG]},\n\tkeywords = {Computer Science - Machine Learning},\n}\n\n","author_short":["Haque, M. R.","Elhabian, S.","Pettine, W. W."],"key":"haque_timesynth_2026","id":"haque_timesynth_2026","bibbaseid":"haque-elhabian-pettine-timesynthatemporalfidelityframeworkforhealthsignaldigitaltwins-2026","role":"author","urls":{"Paper":"http://arxiv.org/abs/2607.00431"},"keyword":["Computer Science - Machine Learning"],"metadata":{"authorlinks":{}}},"bibtype":"misc","biburl":"https://api.zotero.org/users/1528358/collections/DH4R6ZWA/items?key=L8lztpUgybScRoWSW76OhIE0&format=bibtex&limit=100","dataSources":["hoqThwEepw4WmWudJ"],"keywords":["computer science - machine learning"],"search_terms":["timesynth","temporal","fidelity","framework","health","signal","digital","twins","haque","elhabian","pettine"],"title":"Timesynth: A Temporal Fidelity Framework for Health Signal Digital Twins","year":2026}