Parallel computing: how best to retrieve outputs as jobs complete?

I am doing the following: run a set of jobs on a cluster, collect their outputs, process, [downstream code].

My question is how to improve my implementation of the "collect their outputs" step. The only way I've found to do this involves waiting and periodically checking for completed jobs. Is there a better way? Can jobs signal their completion and trigger processing of their output?

My simplified code is below. Note that it uses an LSF scheduler instead of the Matlab Job Scheduler (MJS), which means that some of the method names are different (.destroy instead of .delete, .getAllOutputArguments instead of .fetchOutputs).

MATLAB code
%construct & submit jobs
jobArray{i} = cell(1,numJobs)
for i=1:numJobs
 jobArray{i} = createJob(schedulerObject);
 createTask(jobArray{i},@somFunct,numOutputs,someInputs);
 submit(jobArray{i});
end
  
  
%collect outputs, delete jobs
outputs = cell(size(jobArray));
jobsDeleted = 0; 
numJobs = length(jobArray);
jobsNotDone = ~cellfun('isempty',jobArray);
pause on
while jobsDeleted < numJobs

    %wait to check for completed jobs
    pause(someTime);  

    %check for completed jobs
    for i = find(jobsNotDone(:))'
        if strcmp(jobArray{i}.State,'finished')

            %get output
            thisOutput = jobArray{i}.getAllOutputArguments;

            %process output
            % [...]

            %record output
            output{i} = thisOutputProcessed;

            %clean up
            jobsNotDone(i) = false;
            jobArray{i}.destroy;
            jobsDeleted = jobsDeleted + 1;
        elseif strcmp(jobArray{i}.State,'failed') 
            %print to error file
            % [...]

            jobsNotDone(i) = false;
            jobsDeleted = jobsDeleted + 1;
        end
    end
end
pause off

Thank you.