Methods and devices in performing a vision testing procedure on a person

Patent metadata

Field	Value
Publication number	US-12572802-B2
Application number	US-202418660605-A
Country	US
Kind code	B2
Filing date	May 10, 2024
Priority date	Aug 18, 2022
Publication date	Mar 10, 2026
Grant date	Mar 10, 2026

Abstract

A computer-implemented method for training of a machine learning model for determining a confidence value during at least one test cycle of a vision testing procedure is disclosed. The confidence value is designated to determine at least one action in at least one subsequent test cycle of the vision testing procedure. Further, a trained machine learning model, a computer program having instructions for training of the machine learning model and a training apparatus are disclosed. Additionally, a computer-implemented method for performing the vision testing procedure on a person, a computer program having instructions for performing the vision testing procedure, a vision test apparatus, and a method for producing a geometrical model of at least one spectacle lens for manufacturing of at least one spectacle lens are disclosed.

First claim

The invention claimed is:

1 . A computer-implemented method for training of a machine learning model for generating a trained machine learning model to determine a confidence value during a test cycle of a vision testing procedure, wherein the computer-implemented method is implemented as at least one computer program, wherein the confidence value is designated for determining an action in a subsequent test cycle of the vision testing procedure; the method comprising the following steps, which are performed by using the computer program:
a) providing training data including
first information about a behavior of a person during the test cycle of the vision testing procedure, wherein the behavior of the person as provided by the first information is recorded from an observation of the person during a time interval after the person is queried to solve a task by providing an expected response; and
second information about a known confidence value;

b) determining the confidence value of the person, wherein the confidence value is a measure of a confidence level of the person at providing the expected response, by analyzing the first information, using the machine learning model, and determining a deviation between the determined confidence value and the known confidence value as provided by the second information; and
c) adjusting the machine learning model in order to minimize the deviation between the determined confidence value and the known confidence value by changing a parameter of the machine learning model,
wherein the steps a) to c) are repeated until a determination criterion is met that, when met, leads to a termination of the training of the machine learning model, wherein the parameter of the machine learning model that resulted by determining a minimal deviation is used in the trained machine learning model,
wherein the confidence value indicates how sure the person is in giving the expected response,
wherein the vision testing procedure is performed for determining a condition of at least one eye of the person, wherein the condition is a performance indicator of the vision of the person,
wherein the first information about the behavior of the person includes information about at least one of:
 at least one expression in a body of the person; or
 at least one expression in a voice of the person,
wherein the confidence value is selected from:
 at least one discrete scale; or
 at least one continuous scale.

2 . The method according to claim 1 , wherein the machine learning model is further trained for generating the trained machine learning model to further determine a correctness value, wherein the correctness value is a measure of a probability that a response provided by the person is the expected response,
wherein the first information further includes information about the response provided by the person, wherein the training data further includes third information about a known correctness value, wherein the correctness value is determined by analyzing the first information using the machine learning model, wherein a first further deviation is determined between the determined correctness value and the known correctness value as provided by the third information, wherein the parameters of the machine learning model are changed when at least a portion of the training data is analyzed by the machine learning model, wherein the further training is repeated until a first further determination criterion is met that, when met, leads to the termination of the further training of the machine learning model, and wherein the parameters of the machine learning model that resulted by determining a minimal first further deviation are used in the trained machine learning model.

3 . The method according to claim 1 , wherein the machine learning model is further trained for generating the trained machine learning model to further determine the action in the subsequent test cycle of the vision testing procedure, wherein the training data further comprises:
fourth information about a known action,
wherein the action is determined by analyzing the determined confidence value by using the machine learning model, wherein a second further deviation is determined between the determined action and the action provided by the fourth information, wherein the parameters of the machine learning model are changed when at least a portion of the training data is analyzed by the machine learning model, wherein the further training is repeated until a second further determination criterion is met that, when met, leads to the termination of the further training of the machine learning model, and wherein parameters of the machine learning model that resulted by determining a minimal second further deviation are used in the trained machine learning model.

4 . The trained machine learning model, wherein the machine learning model has been trained by the computer-implemented method for training of the machine learning model according to claim 1 .

5 . The computer-implemented method for performing the vision testing procedure on the person, wherein the vision testing procedure includes at least two subsequent test cycles, wherein a subsequent test cycle comprises at least the following steps:
d) presenting the task to a person querying the person to provide an expected response by using a presentation device; e) recording measurement data by using a recording device, including
the first information about the behavior of the person during the test cycle of the vision testing procedure, wherein the behavior of the person as provided by the first information is recorded from an observation of the person during the time interval after the person is queried to solve the task by providing the expected response;
f) determining the confidence value of the person by analyzing the first information using the machine learning model being trained by the method according to claim 1 by using the processing device; and g) determining the action in the subsequent test cycle of the vision testing procedure by considering the determined confidence value.

6 . The method according to claim 1 , wherein the performance indicator of the vision of the person is an impairment of the person being expressed by a visual parameter of the at least one eye of the person, and wherein the visual parameter is selected from at least one of a refractive error or a visual performance of the at least one eye of the person.

7 . The method according to claim 1 , wherein solving the task is performed by the person using an assistive device, wherein the assistive device has a degree of influencing the ability of the person to solve the task, wherein the assistive device is configured to counteract an impairment of the person impeding the person to solve the task.

8 . The method according to claim 1 , wherein the determined action is selected from at least one of:
giving a feedback to the person whether the provided response was the expected response, optionally before presenting the task in the subsequent test cycle; querying an indication on an estimate of at least one certainty from the person, optionally before presenting the at least one task in the at least one subsequent test cycle; changing a time-to-answer in the at least one subsequent test cycle; maintaining a presented task and presenting the presented task again in the subsequent test cycle; changing the presented task and presenting a further task in the subsequent test cycle which differs from the presented task; maintaining a used assistive device and using the used assistive device again in the subsequent test cycle; changing the used assistive device and using a further assistive device in the subsequen

Assignees

• Zeiss Carl Vision Int Gmbh

Inventors

• Leube Alexander
• Thom Dennis

Classifications

• A61B3/028

  for testing visual acuity; for determination of refraction, e.g. phoropters · CPC title

• G06N20/00

  Machine learning · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G16H50/20

  for computer-aided diagnosis, e.g. based on medical expert systems · CPC title