To investigate the customer reported issue of very slow processing of a 7-day holter study, the customer provided philips customer support with the patient¿s raw data file for review and analysis.Philips remote clinical support analyzed the data file and found that, of the 3 channels of data, channels 2 and 3 mainly consisted of artifact (unrecognizable area/beat [of unknown morphology]) with very limited useful information.As a result of the artifact on channels 2 and 3, the software would have a very difficult time analyzing the data from these two channels.If the user still wanted to use channels 2 and 3 for analysis, they have to make many corrections to the data before completing the report, which is what the customer experienced.According to philips remote clinical support, the patient was disconnected periodically throughout the 7 days - the scan of the patient began on (b)(6) 2021 and was completed on (b)(6) 2021.Analysis of the patient data by the customer began on (b)(6) 2021 and, according to the customer, it took two weeks to analyze the data.Further information about the patient, cause of death and relationship (if any) of the patient death to the reported lengthy analysis time was requested, but the customer declined to provide any further information citing patient confidentiality regulations.In this case, the customer used a method of scanning that allows the system to analyze the recording without interaction from the user (retro mode).With this method, if there is a channel that has too much artifact or if the algorithm picks up a beat incorrectly instead of correcting it on the spot, the ¿mislabeled beat/event¿ continues throughout the recording.When this occurs, the user is required to go into the events at the end of the retro mode analysis and manually correct any issues with the data analysis.According to philips remote clinical support, this is not a preferred method of scanning especially on a artifactual study or a difficult rhythm.Users prefer this method because it is fast in the beginning but then the corrections can take some time to complete.In the case of the data related to this particular patient, the user should turn off channels 2 and 3 for analysis and then use the retro scanning mode.Philips remote clinical support stated that, after reviewing the patient¿s recording, the slowness of processing claimed by the customer was most likely due to the poor quality of the patient data and the patient¿s leads becoming disconnected almost daily.Typically, with this quality of data, the test would be required to be re-performed.To analyze data with large amounts of artifact and patient lead disconnection, philips remote clinical support suggested to the customer that they utilize page mode, which displays the ecg one page at a time.Page mode is a prospective mode that allows the user to supervise the analysis by viewing the ecg chronologically and fine-tuning the arrhythmia processor online for error-free results.Page mode gives constant feedback on the performance of the arrhythmia algorithm, with complexes identified by labels and color highlighting.In this mode, the user can classify unknown beats as specific types of beats through the learn process.The mode that the customer was using for this patient¿s data - retro mode - does not allow for user intervention unless the user manually stops the scanning process.With regards to page mode, the software¿s ifu (page 2-10) states: ¿optimally, you can use page mode for the first hour of patient data, then switch to quick scan or retro.This approach reduces the amount of time you have to spend editing.¿ with regards to retro mode, the software¿s ifu states: ¿this scanning method is completely automatic.It is appropriate for patients who have relatively normal ecgs and artifact-free recordings.¿ philips remote clinical support attempted to contact the customer after providing the above guidance and did not receive a response from the customer.According to philips remote clinical support, they advised the customer of the need to have additional discussion about patient skin prep since this particular patient¿s recording was full of artifact and loss of leads for long periods of time.According to philips remote clinical support, better skin prep, good quality electrodes and sending home extra electrodes with the patient may have helped reduce the artifact.Additionally, the analysis time would have been reduced by turning off the channels with the artifact or erasing any periods of time full of artifact while paging through the exam.The software¿s ifu (page 7-11, noise) provides additional suggestions for troubleshooting to help fix noise problems.The investigation did not confirm that a malfunction of the philips software occurred.Poor quality of the patient data and the patient¿s leads becoming disconnected almost daily led to large amounts of artifact in the patient data file being analyzed.With the poor-quality data and large amount of artifacts (e.G., 8000 ¿events¿), a lengthy amount of time would be spent reviewing, analyzing and editing the data before generating the holter report.This is normal behavior of the holter system.In addition, the method chosen for the data analysis (retro mode) is not optimal per the software¿s ifu given the large amount of artifact on 2 of the 3 data channels.The investigation was unable to conclude that the holter system had malfunctioned during the processing of the patient¿s data or that the holter system was a causal or contributing factor in the patient¿s death.
|