US 11,941,544 B2
Generating and applying robust dose prediction models
Perttu Niemela, Espoo (FI); Jari Lindberg, Helsinki (FI); Tuomas Jyske, Vantaa (FI); Maria Cordero Marcos, Espoo (FI); and Esa Kuusela, Espoo (FI)
Assigned to SIEMENS HEALTHINEERS INTERNATIONAL AG, Steinhausen (CH)
Filed by Varian Medical Systems International AG., Cham (CH)
Filed on Dec. 22, 2022, as Appl. No. 18/087,756.
Application 18/087,756 is a continuation of application No. 16/795,143, filed on Feb. 19, 2020, granted, now 11,537,912.
Prior Publication US 2023/0130526 A1, Apr. 27, 2023
Int. Cl. A61N 5/10 (2006.01); G06N 5/04 (2023.01); G06N 20/00 (2019.01); G16H 20/40 (2018.01)
CPC G06N 5/04 (2013.01) [A61N 5/1031 (2013.01); G06N 20/00 (2019.01); G16H 20/40 (2018.01); A61N 2005/1041 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
accessing a plurality of treatment plans;
selecting information associated with the plurality of treatment plans to train one or more dose prediction models, wherein the selected information includes nominal values of parameters of the plurality of treatment plans and perturbations of the nominal values;
training one or more dose prediction models based on the selected information including the nominal values of the parameters of the plurality of treatment plans and the perturbations of the nominal values; and
developing one or more new treatment plans based on the one or more dose prediction models.