IP Library Granted Patent US 12,640,241
Granted Patent B2
US 12,640,241 · App. 16/289,450 · Granted May 26, 2026

Systems and methods for determining patient hospitalization risk and treating patients

Inventors: Andrew W. Long (Brookline, MA); Thomas C. Blanchard (Somerville, MA); Len Usvyat (Boston, MA); Hernando G. Garza (Concord, CA); Jodi Conti (Lakeway, TX); Cara S. Gallagher (Leander, TX); Joanna L. Willetts (Framingham, MA); Hao Han (Lexington, MA); Sheetal Chaudhuri (Arlington, MA); Franklin W. Maddux (Lincoln, MA)
Assignee: Fresenius Medical Care Holdings, Inc.
G16H10/60G16H20/40G16H50/30
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 12,640,241
App. No.
16/289,450
Granted
May 26, 2026
Kind
B2
Abstract

A system and method for determining patient hospitalization risk and treating patients is disclosed. The system and method may include extracting patient data from one or more databases corresponding to a pool of patients having end stage renal disease; using a predictive model with the extracted patient data to generate, for each of the patients, a respective expected probability for hospitalization within a predetermined time period; identifying a subset of patients having respective expected probabilities that are higher than other patients in the pool of patients; identifying, for each patient, at least one factor from the patient data that increased the expected probability of hospitalization; and based on the identified factors, determining and executing clinical interventions to lower the probability of hospitalization within the subset of the pool of patients.

Claims (59)

1 . A method, comprising:

extracting, by at least one computing device, historical patient data from one or more databases, wherein the historical patient data corresponds to a pool of patients having end stage renal disease (ESRD);

using, by the at least one computing device, a predictive machine learning model with the extracted historical patient data corresponding to the pool of ESRD patients, for each ESRD patient in the pool of ESRD patients, to generate a risk score indicating an expected probability of hospitalization within a predetermined time period and to identify a plurality of factors that contributed to the risk score;

identifying, by the at least one computing device, a subset of the pool of ESRD patients having respective expected probabilities of hospitalization that are higher than other ESRD patients in the pool of ESRD patients;

determining, by the at least one computing device, based on the extracted historical patient data corresponding to the pool of ESRD patients, for each ESRD patient of the subset of ESRD patients, at least one prominent factor from the plurality of factors that contributed to the risk score, wherein the at least one prominent factor is configured to support a determination of patient interventions to impact the expected probability of hospitalization, and wherein the at least one prominent factor for each ESRD patient of the pool of ESRD patients is identified using Shapley additive explanations; and

determining and executing, by the at least one computing device, at least one clinical intervention for at least one ESRD patient of the pool of ESRD patients, wherein the at least one clinical intervention is configured to target the at least one prominent factor to lower the expected probability of hospitalization for the at least one ESRD patient, and wherein the at least one clinical intervention comprises at least one of:

(a) administering one or more dialysis treatments in addition to a patient's existing dialysis schedule,

(b) extending a patient's dialysis treatment time,

(c) adjusting a patient's target weight for a dialysis treatment,

(d) adjusting a dialysate sodium concentration for a patient's dialysis treatment, or

(e) adjusting a patient's blood pressure medication.

2 . The method of claim 1 , wherein the predetermined time period is 7 days or less.

3 . The method of claim 1 , wherein the extracted historical patient data corresponding to the pool of ESRD patients comprises a patient demographics.

4 . The method of claim 3 , wherein the patient demographics comprise one or more of: date of birth, date of first dialysis, gender, race, ethnicity, or marital status.

5 . The method of claim 1 , wherein the extracted historical patient data corresponding to the pool of ESRD patients comprises laboratory values, and wherein laboratory values comprise one or more of: hemoglobin level or albumin level.

6 . The method of claim 5 , wherein the laboratory values include one or more of: an average over a specified time period, a maximum value, a minimum value, a value spike, a value dip, or trending values.

7 . The method of claim 1 , wherein the extracted historical patient data corresponding to the pool of ESRD patients comprises treatment data, wherein the treatment data comprises vitals, and wherein the vitals include one or more of: an average over a specified time period, a maximum value, a minimum value, a value spike, a value dip, or trending values.

8 . The method of claim 1 , wherein the extracted historical patient data corresponding to the pool of ESRD patients comprises a comprehensive assessment, wherein the comprehensive assessment comprises one or more of: recent hospitalization history, a history of a missed appointments, notes, complaints, medical professional assessments, or delivered medications.

9 . The method of claim 1 , further comprising:

generating a report that ranks the pool of ESRD patients according to their respective expected probabilities of hospitalization, wherein the report also provides the at least one prominent factor for each respective ESRD patient.

10 . The method of claim 9 , further comprising:

providing the generated report to one or more health care providers.

11 . The method of claim 1 , further comprising:

transmitting an automated alert to one or more health care providers, based on the expected probabilities of hospitalization.

12 . The method of claim 1 , wherein the pool of ESRD patients are patients of an ESRD Seamless Care Organization (ESCO).

13 . A system, comprising:

one or more databases configured to store historical patient data, wherein the historical patient data corresponds to a pool of patients having end stage renal disease (ESRD); and

at least one computing device configured to:

extract the historical patient data from the one or more databases;

use a predictive machine learning model with the extracted historical patient data corresponding to the pool of ESRD patients, for each ESRD patient in the pool of ESRD patients, to generate a risk score indicating an expected probability of hospitalization within a predetermined time period and to identify a plurality of factors that contributed to the risk score;

identify a subset of the pool of ESRD patients having respective expected probabilities of hospitalization that are higher than other ESRD patients in the pool of ESRD patients;

determine, based on the extracted historical patient data corresponding to the pool of ESRD patients, for each ESRD patient of the subset of ESRD patients, at least one prominent factor from the plurality of factors that contributed to the risk score, wherein the at least one prominent factor is configured to support a determination of patient interventions to impact the expected probability of hospitalization, and wherein the at least one prominent factor for each ESRD patient of the pool of ESRD patients is identified using Shapley additive explanations; and

determine and execute at least one clinical intervention for at least one ESRD patient of the pool of ESRD patients, wherein the at least one clinical intervention is configured to target the at least one prominent factor to lower the expected probability of hospitalization for the at least one ESRD patient, and wherein the at least one clinical intervention comprises at least one of:

(a) administering one or more dialysis treatments in addition to a patient's existing dialysis schedule,

(b) extending a patient's dialysis treatment time,

(c) adjusting a patient's target weight for a dialysis treatment,

(d) adjusting a dialysate sodium concentration for a patient's dialysis treatment, or

(e) adjusting a patient's blood pressure medication.

14 . The system of claim 13 , wherein the predetermined time period is 7 days or less.

15 . The system of claim 13 , wherein the at least one computing device is further configured to:

generate a report that ranks the pool of ESRD patients according to their respective expected probabilities of hospitalization, wherein the report also provides the at least one prominent factor for each respective ESRD patient.

16 . The system of claim 13 , wherein the at least one computing device is further configured to:

transmit an automated alert to one or more health care providers, based on the expected probabilities of hospitalization.

17 . One or more non-transitory computer-readable mediums having processor-executable instructions stored thereon, wherein the processor-executable instructions, when executed, facilitate performance of the following:

extracting, by at least one computing device, historical patient data from one or more databases, wherein the historical patient data corresponds to a pool of patients having end stage renal disease (ESRD);

using, by the at least one computing device, a predictive machine learning model with the extracted historical patient data corresponding to the pool of ESRD patients, for each ESRD patient in the pool of ESRD patients, to generate a risk score indicating an expected probability of hospitalization within a predetermined time period and to identify a plurality of factors that contributed to the risk score;

identifying, by the at least one computing device, a subset of the pool of ESRD patients having respective expected probabilities of hospitalization that are higher than other ESRD patients in the pool of ESRD patients;

determining, by the at least one computing device, based on the extracted historical patient data corresponding to the pool of ESRD patients, for each ESRD patient of the subset of ESRD patients, at least one prominent factor from the plurality of factors that contributed to the risk score, wherein the at least one prominent factor is configured to support a determination of patient interventions to impact the expected probability of hospitalization, and wherein the at least one prominent factor for each ESRD patient of the pool of ESRD patients is identified using Shapley additive explanations; and

determining and executing, by the at least one computing device, at least one clinical intervention for at least one ESRD patient of the pool of ESRD patients, wherein the at least one clinical intervention is configured to target the at least one prominent factor to lower the expected probability of hospitalization for the at least one ESRD patient, and wherein the at least one clinical intervention comprises at least one of:

(a) administering one or more dialysis treatments in addition to a patient's existing dialysis schedule,

(b) extending a patient's dialysis treatment time,

(c) adjusting a patient's target weight for a dialysis treatment,

(d) adjusting a dialysate sodium concentration for a patient's dialysis treatment, or

(e) adjusting a patient's blood pressure medication.

18 . The one or more non-transitory computer-readable mediums of claim 17 , wherein the predetermined time period is 7 days or less.

19 . The one or more non-transitory computer-readable mediums of claim 17 , wherein the processor-executable instructions, when executed, further facilitate performance of the following:

generating a report that ranks the pool of ESRD patients according to their respective expected probabilities of hospitalization, wherein the report also provides the at least one prominent factor for each respective ESRD patient.

20 . The one or more non-transitory computer-readable mediums of claim 17 , wherein the processor-executable instructions, when executed, further facilitate performance of the following:

transmitting an automated alert to one or more health care providers, based on the expected probabilities of hospitalization.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded May 9, 2019
From: LONG, ANDREW W.; BLANCHARD, THOMAS C.; USVYAT, LEN; GARZA, HERNANDO G.; CONTI, JODI; GALLAGHER, CARA S.; WILLETTS, JOANNA L.; HAN, HAO; CHAUDHURI, SHEETAL; MADDUX, FRANKLIN W.
To: FRESENIUS MEDICAL CARE HOLDINGS, INC.
Reel/Frame 049127/0981 →
Continuity (3)
Provisional Application 62637332 · Mar 1, 2018
Provisional Application 62716034 · Aug 8, 2018
Related Publication 20200051674A1 · Feb 13, 2020
References Cited (38)
US 20060167654A1 · Keinan et al. · 2006 [cited by applicant]
US 20060167784A1 · Hoffberg · 2006 [cited by examiner]
US 20100023307A1 · Lee · 2010 [cited by examiner]
US 20130144642A1 · Bessette · 2013 [cited by examiner]
US 20130262357A1 · Amarasingham · 2013 [cited by examiner]
US 20140278546A1 · Op Den Buijs et al. · 2014 [cited by applicant]
US 20150213217A1 · Amarasingham · 2015 [cited by examiner]
US 20150213224A1 · Amarasingham · 2015 [cited by examiner]
US 20150213225A1 · Amarasingham · 2015 [cited by examiner]
US 20160357923A1 · Dong · 2016 [cited by examiner]
US 20170083682A1 · McNutt · 2017 [cited by examiner]
US 20180158552A1 · Liu · 2018 [cited by examiner]
US 20180166174A1 · Lewis · 2018 [cited by examiner]
US 20180374581A1 · Berringer · 2018 [cited by examiner]
US 20190034590A1 · Oren · 2019 [cited by examiner]
US 20190192052A1 · Weffers-Albu · 2019 [cited by examiner]
US 20190213685A1 · Ironside · 2019 [cited by examiner]
US 20190378210A1 · Merrill · 2019 [cited by examiner]
US 20220005041A1 · Chang · 2022 [cited by examiner]
US 20220181027A1 · Ackerman · 2022 [cited by examiner]
CA 3008904A1 · 2017 [cited by applicant]
WO 2017106558A1 · 2017 [cited by applicant]
Titapiccolo et al., “Mining Medical Data to Develop Clinical Decision Making Tools in Hemodialysis,” 2012 IEEE 12th International Conference on Data Mining Workshops, Brussels, Belgium, 2012, pp. 99-106, doi: 10.1109/IC… [cited by examiner]
Cornalba et al., “Building a Normative Decision Support System for Clinical and Operational Risk Management in Hemodialysis,” in IEEE Transactions on Information Technology in Biomedicine, vol. 12, No. 5, pp. 678-686, S… [cited by examiner]
Sun et al. 2006, “Boosting Kernel Models for Regression,” Sixth International Conference on Data Mining (ICDM'06), Hong Kong, China, 2006, pp. 583-591, doi: 10.1109/ICDM.2006.30. [cited by examiner]
Natekin et al. 2013, “Gradient boosting machines, a tutorial.” Frontiers in neurorobotics vol. 7 21. Dec. 4, 2013, doi: 10.3389/fnbot.2013.00021. [cited by examiner]
Diego et al. 2016, “Structured Regression Gradient Boosting,” 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 2016, pp. 1459-1467, doi: 10.1109/CVPR.2016.162. [cited by examiner]
International Search Report and Written Opinion for International application No. PCT/US2019/020129, mailed on May 9, 2019, 17 pages. [cited by applicant]
Wick, J.P. et al, “A Clinical Risk Prediction Tool for 6-Month Mortality After Dialysis Initiation Among Older Adults”, American Journal of Kidney Diseases , 69(5):568-575 (2016). [cited by applicant]
Couchoud, C., et al, “A clinical score to predict 6-month prognosis in elderly patients starting dialysis for end-stage renal disease”, Nephrology Dialysis Transplant 24(5):1553-1561 (2008). [cited by applicant]
Thamer, M., et al., “Predicting Early Death Among Elderly Dialysis Patients: Development and Validation of a Risk Score to Assist Shared Decision making for Dialysis Initiation”, American Journal of Kidney Diseases 66(6… [cited by applicant]
Mingle, D., “Predicting Diabetic Readmission Rates: Moving Beyond Hba1c”, Current Trends in Biomedical Engineering & Biosciences, 7(3):1-11 (2017). [cited by applicant]
Lundberg, S., and Lee, S-I, “A Unified Approach to Interpreting Model Predictions”. Paper presented at the 31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA (Nov. 2017), 10 pages. [cited by applicant]
Lundberg, S., et al., “Consistent Individualized Feature Attribution for Tree Ensembles”, ACM Conference [online] London UK (Jul. 2018) [retrieved on May 14, 2019]. Retrieved from Internet URL: https://groundai.com/proj… [cited by applicant]
Couchoud, et al., Development of a risk stratification algorithm to imporve patient-centered care and decision making for incident elderly patients with end-stage renal disease, Kidney International 88(5): 1178-1186 (20… [cited by applicant]
Mingle, D., “Predicting Diabetic Readmission Rates: Moving Beyond Hba1c”, Current Trends in Biomedical Engineering & Biosciences, vol. 7, No. 3, Aug. 2017, pp. 55-65, DOI: 10.19080/CTBEB.2017.07.555715. [cited by applicant]
Australian Examination Report No. 2, AU Application No. 2019227922, Mar. 17, 2022 (5 pages). [cited by applicant]
Couchoud et al. “Development of a risk stratification algorithm to improve patient-centered care and decision making for incident elderly patients with end-stage renal disease”—Kidney International (2015)—Sep. 2, 2015—p… [cited by applicant]