Page 46 - Read Online

Farinha et al. Mini-invasive Surg 2023;7:38 https://dx.doi.org/10.20517/2574-1225.2023.50 Page 7 of 14

Fifty-eight participants were enrolled in two animals [9,16] , 83 participants in three 3D [17,19,29] , and 42 in one
[27]
VR/AR TM study .

The study participants were medical students, residents, fellows, and attending surgeons. The criteria used
to classify them into “novice”, “intermediate”, and “expert” groups varied between studies [16,17,19,27,29] . For
[19]
example, the definition of “expert”, as a surgeon with > 100 [16,27] or > 150 console cases , was based on the
number of surgical cases completed [16,19,27,29] . The experiences of the different enrolled cohorts varied
considerably, including subjects without any surgical experience [17,27] . Comparisons between two groups
with a clear discrimination of surgical experience (novices and experts) , and three groups without a clear
[29]
difference in experience (novices, intermediates, and experts) [9,16,17,19] were identified.

Photo or video recordings of the surgeon’s performance were collected, and experts were blinded to the
experience level and the surgeon performing the task. The metrics used varied from GEARS [9,19,27] ,
GOALS [16,29] , and clinically relevant outcome measures (CROMS) to different operation-specific metrics,
[19]
namely, time (renal artery clamping [17,19] , tumor excision [9,34] , total operative [9,16] , and console time ),
[19]
[19]
estimated blood loss , preserved renal parenchyma , surgical margin status [16,17,19,29] , maximum gap
[17]
between the two sides of the incision , total split length , and quality of PN (scored on a Likert scale) . In
[29]
[9]
[29]
one animal model, instrument and camera awareness and the precision of instrument action were
[27]
subjectively scored using a Likert scale . Built-in algorithm software metrics were used in one VR TM,
scoring instrument collisions, instrument time out of view, excessive instrument force, economy of motion,
[27]
time to task completion, and incorrect answers [Table 3].
Concurrent validity
One AR/VR simulator study compared the performance of experts on a virtual and an in vivo porcine
renorrhaphy task. It was found to have equal realism and high usefulness for teaching anatomy, procedural
steps, and training technical skills of residents and fellows, although less so for experienced robotic surgeons
new to RAPN .
[27]
Kane’s framework
Following Kane’s framework of the validation process, focusing on decisions and consequences, the
[18]
fragilities of the analyzed studies become more obvious. The proposed use of different models varies from
developing and testing them to evaluating distinct levels of validation [9,14,16-29] . The type of scoring used is
based on the timing of various steps of the emulated procedure and/or using Likert scales, such as GEARS
or GOALS [9,13,14,16-29,31] .

None of the studies generalized the test results to other tasks. Several authors report their models as realistic
and useful training tools for residents and fellows, although they are usually not considered highly beneficial
for training consultants [9,14,16-29,31] . The implications of using diverse models differ across studies. Generally
considered an effective surgical education/training tool to learn key steps of PN and develop advanced
laparoscopic/robotic skills, they are associated with fewer logistic concerns. This is due to their lack of
necessity for dedicated teaching robots or wet/laboratory facilities [Table 4].

DISCUSSION
The aviation industry established the safety benefit of training on simulators many decades ago , inspiring
[35]
surgeons to pursue their training in the laboratory before entering the operating room [36,37] . Skills acquired