Systems and methods for improving user experience during damage appraisal

A method comprising;

• receiving, by an appraisal management computing apparatus, a plurality of damage evidence files each depicting one or more damaged parts of a vehicle damaged during an adverse incident;
• determining, by the appraisal management computing apparatus, a likelihood of relevance of each damage evidence file to an appraisal of an individual damaged part depicted in the file by using a first machine learning algorithm trained on historic relevance data comprising a plurality of damage evidence files associated with corresponding damaged parts previously identified as relevant;
• presenting, in a graphical user interface (GUI) of an appraisal application associated with a user computing device, a set of relevant damage evidence files for each damaged part of the vehicle, each of the damage evidence files having the likelihood of relevance exceeding a threshold;
• generating a first set of training data comprising previously determined repair recommendations for restoring damaged parts;
• training a second machine learning algorithm in a first stage with the first set of training data;
• responsive to a user selection comprising an individual damaged part, presenting a repair recommendation for restoring the damaged part by using the second machine learning algorithm;
• responsive to the user selection comprising a rejection of the recommendation, generating a second set of training data based on corrections provided by the user, and training the second machine learning algorithm in a second stage with the second set of training data;
• wherein each of the sets of relevant damage evidence associated with each of the damaged parts is presented upon receiving the user selection comprising a corresponding damaged part of the vehicle;
• and
• responsive to the user selection comprising an acceptance of the recommendation, generating a repair estimate line.

A non-transitory machine-readable storage medium encoded with instructions executable by one or more hardware processors of a computing component, the machine-readable storage medium comprising instructions to cause the one or more hardware processors to perform operations comprising:

• receive a plurality of damage evidence files each depicting one or more damaged parts of a vehicle damaged during an adverse incident;
• determine a likelihood of relevance of each damage evidence file to an appraisal of an individual damaged part depicted in the file by using a first machine learning algorithm trained on historic relevance data comprising a plurality of damage evidence files associated with corresponding damaged parts previously identified as relevant;
• present, in a graphical user interface (GUI) of an appraisal application associated with a user computing device, a set of relevant damage evidence files for each damaged part of the vehicle, each of the damage evidence files having the likelihood of relevance exceeding a threshold value;
• generate a first set of training data comprising previously determined repair recommendations for restoring damaged parts;
• train a second machine learning algorithm in a first stage with the first set of training data;
• responsive to a user selection comprising an individual damaged part, present a repair recommendation for restoring the damaged part by using the second machine learning algorithm;
• responsive to the user selection comprising a rejection of the recommendation, generate a second set of training data based on corrections provided by the user, and train the second machine learning algorithm in a second stage with the second set of training data;
• wherein each of the sets of relevant damage evidence files associated with each of the damaged parts is presented upon receiving the user selection comprising a corresponding damaged part of the vehicle;
• and
• responsive to the user selection comprising an acceptance of the recommendation, generate a repair estimate line.