IP Library › Granted Patent US 12,725,682
Granted Patent B2
US 12,725,682 · App. 18/450,160 · Granted Sep 1, 2026

Systems, methods, and articles for imputing directed temporal measurements

Inventors: Samuel Peter Heilbroner (San Francisco, CA); Riccardo Miotto (New York, NY); Dany Michael Haddad (Austin, TX)
Assignee: Tempus Labs, Inc.
G16H10/60
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,725,682
App. No.
18/450,160
Granted
Sep 1, 2026
Kind
B2
Abstract

The present disclosure relates to predicting a data element in an electronic health record (EHR) for a subject using a trained machine learning model including an attention module. An example method includes obtaining a query for the prediction of the data element, obtaining a plurality of observations about the subject, processing the query and observations with the trained machine learning model having an attention module to generate a prediction of the subject characteristic, and providing the prediction of the data element as an output.

Claims (55)

1 . A computer-implemented method for imputing a value associated with a subject within an electronic health record (EHR) system, the method comprising:

receiving, by a computing device and from a second computing device, a request to impute the value associated with the subject at a temporal instance;

retrieving, by the computing device, a subset of data associated with the subject from the EHR system, the subset of data comprising a plurality of stored values associated with one or more temporal instances;

providing, by the computing device, the temporal instance indicated in the request and the subset of data to a trained artificial intelligence engine, the trained artificial intelligence engine configured to perform actions, comprising:

determining, by the computing device, relationships between the stored values by:

calculating a set of scores for multiple subsets of features of the stored values, wherein the set of scores represents interdependencies between the stored values;

determining, by the computing device, time weights based on a temporal proximity of the one or more temporal instances of the stored values relative to the temporal instance of the value being imputed;

adjusting, by the computing device, the calculated scores based on the time weights to generate time-adjusted scores;

generating, by the computing device, an imputed value based on a combination of the stored values and the time-adjusted scores, wherein the imputed value corresponds to at least one lab test value;

providing, by the computing device, the imputed value to the second computing device in response to the request;

generating, by the computing device, one or more visualizations that include a heatmap indicating different predictive relevance of a plurality of values including the imputed value; and

displaying, by the computing device, one or more predicted health conditions for the subject based on the one or more visualizations.

2 . The method of claim 1 , wherein the subset of data are associated with the subject.

3 . The method of claim 1 , wherein the request includes a unit of measure.

4 . The method of claim 1 , wherein determining the time weights based on the temporal proximity of the one or more temporal instances of the stored values relative to the temporal instance of the value being imputed comprises applying a time decay function to the calculated scores that is dependent on differences between the temporal instance of the value being imputed and the one or more temporal instances of the stored values in the EHR.

5 . The method of claim 4 , wherein the time decay function comprises an exponential time decay function or a linear time decay function.

6 . The method of claim 1 , wherein adjusting the calculated scores based on the time weights comprises applying more weight to stored values that are relatively nearer in time to the temporal instance of the value being imputed.

7 . The method of claim 1 , wherein determining the relationships between the stored values in the EHR comprises use of a multi-head attention module.

8 . The method of claim 1 , wherein generating the imputed value includes:

applying a weight matrix to the time-adjusted scores to combine them into a single representation; and

processing the single representation using a classifier to generate the imputed value.

9 . The method of claim 8 , wherein applying the weight matrix to the time-adjusted scores comprises using a Hadamard product module.

10 . The method of claim 8 , wherein processing the single representation to generate the imputed value comprises using a multilayer perceptron module.

11 . The method of claim 1 , further comprising predicting an occurrence of an adverse event based on the imputed value.

12 . The method of claim 1 , further comprising assessing a predicted eligibility for a clinical trial based on the imputed value.

13 . The method of claim 1 , further comprising predicting a gap in care based on the imputed value.

14 . The method of claim 1 , wherein the stored values include at least one prior lab test result.

15 . The method of claim 1 , wherein the stored values include at least one prior clinical assessment result.

16 . The method of claim 1 , wherein the temporal instance of the value being imputed comprises a date.

17 . A computing system for imputing a value associated with a subject within a structured electronic health record (EHR) system, the computing system comprising:

one or more processors; and

one or more non-transitory computer-readable media collectively storing instructions that, when collectively executed by the one or more processors, cause the one or more processors to perform actions, the actions comprising:

receiving a request to impute the value associated with the subject at a temporal instance;

retrieving a subset of data associated with the subject from the EHR system, the subset of data comprising a plurality of stored values associated with one or more temporal instances;

providing the temporal instance indicated in the request and the subset of data to a trained artificial intelligence engine, the trained artificial intelligence engine configured to perform actions, comprising:

determining relationships between the stored values by:

calculating a set of scores for multiple subsets of features of the stored values, wherein the set of scores represents interdependencies between the stored values;

determining time weights based on a temporal proximity of the one or more temporal instances of the stored values relative to temporal instance of the value being imputed;

adjusting the calculated scores based on the time weights to generate time-adjusted scores;

generating an imputed value based on a combination of the stored values and the time-adjusted scores, wherein the imputed value corresponds to at least one lab test value;

providing the imputed value in response to the request;

generating one or more visualizations that include different shadings indicating different predictive relevance of a plurality of values including the imputed value; and

presenting one or more predicted health conditions for the subject based on the one or more visualizations.

18 . One or more non-transitory computer readable media collectively storing a computer program thereon, the program, when collectively executed by one or more processors, implements operations for imputing a value associated with a subject within a structured electronic health record (EHR) system, the operations comprising:

receiving a request to impute the value associated with the subject at a temporal instance;

retrieving a subset of data associated with the subject from the EHR system, the subset of data comprising a plurality of stored values associated with one or more temporal instances;

providing the temporal instance indicated in the request and the subset of data to a trained artificial intelligence engine, the trained artificial intelligence engine configured to perform actions, comprising:

determining relationships between the stored values by:

calculating a set of scores for multiple subsets of features of the stored values, wherein the set of scores represents interdependencies between the stored values;

determining time weights based on a temporal proximity of the one or more temporal instances of the stored values relative to the temporal instance of the value being imputed;

adjusting the calculated scores based on the time weights to generate time-adjusted scores;

generating an imputed value based on a combination of the stored values and the time-adjusted scores, wherein the imputed value corresponds to at least one lab test value;

providing the imputed value in response to the request;

generating one or more visualizations that include different colors indicating different predictive relevance of a plurality of values including the imputed value; and

causing presentation of one or more predicted health conditions for the subject based on the one or more visualizations.

Assignments (4)
RELEASE OF SECURITY INTEREST Recorded May 14, 2026
From: ARES CAPITAL CORPORATION, AS COLLATERAL AGENT
To: TEMPUS AI, INC. (F/K/A TEMPUS LABS, INC.)
Reel/Frame 075577/0513 →
SECURITY INTEREST Recorded Jun 2, 2025
From: TEMPUS AI, INC.
To: ARES CAPITAL CORPORATION, AS COLLATERAL AGENT
Reel/Frame 071468/0107 →
CHANGE OF NAME Recorded Feb 15, 2024
From: TEMPUS LABS, INC.
To: TEMPUS AI, INC.
Reel/Frame 066600/0899 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Dec 1, 2023
From: HEILBRONER, SAMUEL PETER; MIOTTO, RICCARDO; HADDAD, DANY MICHAEL
To: TEMPUS LABS, INC.
Reel/Frame 065740/0001 →
Continuity (1)
Related Publication 20250061989A1 · Feb 20, 2025
References Cited (59)
US 11177041B1 · Sutton · 2021 [cited by examiner]
US 11475988B1 · Burkart · 2022 [cited by examiner]
US 20080133275A1 · Haug · 2008 [cited by examiner]
US 20110105852A1 · Morris · 2011 [cited by examiner]
US 20170134493A1 · Chang · 2017 [cited by examiner]
US 20190361934A1 · Rogynskyy · 2019 [cited by examiner]
US 20220058663A1 · Cui · 2022 [cited by examiner]
US 20230238145A1 · Patel · 2023 [cited by examiner]
US 20230245777A1 · Foschini · 2023 [cited by examiner]
US 20250201360A1 · Fernandes · 2025 [cited by examiner]
Li, J., Yan, X.S., Chaudhary, D. et al. Imputation of missing values for electronic health record laboratory data. npj Digit. Med. 4, 147 (2021). (Year: 2021). [cited by examiner]
Goldberger, et al., “PhysioBank, PhysioToolkit, and PhysioNet: Components of a New Research Resource for Complex Physiologic Signals,” [cited by applicant]
Heilbroner et al., “Life: A Deep Learning Framework for Laboratory Data Imputation in Electronic Health Records,” Retrieved from https://www.medrxiv.org/content/10.1101/2023.10.31.23297843v1.full, posted Nov. 2, 2023. (… [cited by applicant]
International Search Report and Written Opinion, dated Mar. 12, 2024, for International Patent Application No. PCT/US2023/030239. (24 pages). [cited by applicant]
Ni et al., “MBGAN: An improved generative adversarial network with multi-head self-attention and bidirectional RNN for time series imputation,” [cited by applicant]
Oh et al., “STING: Self-attention based Time-series Imputation Networks using GAN,” Cornell University Library, published online Sep. 22, 2022. (10 pages). [cited by applicant]
Van Buuren et al., “mice: Multivariate Imputation by Chained Equations in R,” [cited by applicant]
Angelopoulos et al., “A Gentle Introduction to Conformal Prediction and Distribution-Free Uncertainty Quantification,” arXiv Preprint at http://arxiv.org/abs/2107.07511, Dec. 2022. (51 pages). [cited by applicant]
Ayilara et al., “Impact of Missing Data on Bias and Precision When Estimating Change in Patient-Reported Outcomes from a Clinical Registry,” [cited by applicant]
Bodenreider et al., “Recent Developments in Clinical Terminologies—SNOMED CT, LOINC, and RxNorm,” [cited by applicant]
Cao et al., “BRITS: Bidirectional Recurrent Imputation for Time Series,” [cited by applicant]
Che et al., “Recurrent Neural Networks for Multivariate Time Series with Missing Values,” [cited by applicant]
De Freitas et al., “Phe2vec: Automated Disease Phenotyping Based on Unsupervised Embeddings from Electronic Health Records,” [cited by applicant]
Dean et al., “Large Scale Distributed Deep Networks,” NIPS'12: [cited by applicant]
Devlin et al., “BERT: Pre-Training of Deep Bidirectional Transformers for Language Understanding,” ar Xiv Preprint at https://doi.org/10.48550/arXiv.1810.04805, May 2019. (16 pages). [cited by applicant]
Falconer et al., “Systematic Review of Predictive Risk Models for Adverse Drug Events in Hospitalized Patients,” [cited by applicant]
Guo et al., “Characterizing the Limitations of Using Diagnosis Codes in the Context of Machine Learning for Healthcare,” [cited by applicant]
Heilbroner et al., “Deep Learning in Medicine,” [cited by applicant]
Horton et al., “Much Ado About Nothing: A Comparison of Missing Data Methods and Software to Fit Incomplete Data Regression Models,” [cited by applicant]
Jensen et al., “Mining Electronic Health Records: Towards Better Research Applications and Clinical Care,” [cited by applicant]
Johnson et al., “Data Descriptor: MIMIC-III, a Freely Accessible Critical Care Database,” [cited by applicant]
Johnson et al., “MIMIC-III Clinical Database (v1.4),” [cited by applicant]
Krishnan et al., “Self-Supervised Learning in Medicine and Healthcare,” [cited by applicant]
Landi et al., “Deep Representation Learning of Electronic Health Records to Unlock Patient Stratification at Scale,” [cited by applicant]
LeCun et al., “Deep Learning,” [cited by applicant]
Li et al., “Generic Medical Concept Embedding and Time Decay for Diverse Patient Outcome Prediction Tasks,” [cited by applicant]
Li et al., “Imputation of Missing Values for Electronic Health Record Laboratory Data,” [cited by applicant]
Luo et al., “3D-MICE: Integration of Cross-Sectional and Longitudinal Imputation for Multi-Analyte Longitudinal Clinical Data,” [cited by applicant]
Luo, Yuan, “Evaluating the State of the Art in Missing Data Imputation for Clinical Data,” [cited by applicant]
Ma et al., “Identifiable Generative Models for Missing Not at Random Data Imputation,” [cited by applicant]
McCord et al., “Using Electronic Health Records for Clinical Trials: Where do we Stand and Where Can we Go?,” [cited by applicant]
Miotto et al., “Deep Learning for Healthcare: Review, Opportunities and Challenges,” [cited by applicant]
Pang et al., “CEHR-BERT: Incorporating Temporal Information from Structured EHR Data to Improve Prediction Tasks,” [cited by applicant]
Rasmy et al., “Med-BERT: Pretrained Contextualized Embeddings on Large-Scale Structured Electronic Health Records for Disease Prediction,” [cited by applicant]
Schadow et al., “The Unified Code for Units of Measure,” UCUM / UCUM Specification, Version: 2.1, pp. 1-84, https://ucum.org/trac, access date Nov. 29, 2021. [cited by applicant]
Scikit-Learn, 1.3.2 Documentation, “6.3. Preprocessing Data,” pp. 1-16, https://scikit-learn.org/stable/modules/preprocessing.html, download date Nov. 13, 2023. [cited by applicant]
Shao et al., “Hybrid Value-Aware Transformer Architecture for Joint Learning from Longitudinal and Non-Longitudinal Clinical Data,” medRxiv preprint at https://www.medrxiv.org/content/10.1101/2023.03.09.23287046v1, Mar.… [cited by applicant]
Shickel et al., Deep Ehr: A Survey of Recent Advances in Deep Learning Techniques for Electronic Health Record (EHR) Analysis, [cited by applicant]
Shulman, David, “Optimization Methods in Deep Learning: A Comprehensive Overview,” arXiv Preprint at https://doi.org/10.48550/arXiv.2302.09566, Feb. 2023. (5 pages). [cited by applicant]
Sun et al., “A Review of Deep Learning Methods for Irregularly Sampled Medical Time Series Data,” ArXiv201012493, Cs Stat, pp. 1-20, Preprint at http://arxiv.org/abs/2010.12493, Oct. 2020. [cited by applicant]
Tomasev et al., “Developing Continuous Risk Models for Adverse Event Prediction in Electronic Health Records using Deep Learning,” [cited by applicant]
Van Buuren et al., “Mice: Multivariate Imputation by Chained Equations in R,” Journal of Statistical Software 45(3): 1-67, Dec. 2011. [cited by applicant]
Van Ginkel et al., “Rebutting Existing Misconceptions About Multiple Imputation as a Method for Handling Missing Data,” [cited by applicant]
Vaswani et al., “Attention Is All You Need,” [cited by applicant]
Xiao et al., “Opportunities and Challenges in Developing Deep Learning Models Using Electronic Health Records Data: a Systematic Review,” [cited by applicant]
Xu et al., “A Multi-Directional Approach for Missing Value Estimation in Multivariate Time Series Clinical Data,” [cited by applicant]
Yasrebi-de Kom et al., “Electronic Health Record-Based Prediction Models for in-Hospital Adverse Drug Event Diagnosis or Prognosis: a Systematic Review,” [cited by applicant]
Yoon et al., “Estimating Missing Data in Temporal Data Streams Using Multi-Directional Recurrent Neural Networks,” [cited by applicant]
Yoon et al., “Multi-Directional Recurrent Neural Networks: A Novel Method for Estimating Missing Data,” [cited by applicant]