This script exports EliteHRV raw data to a .json file:

    ./elitehrv-sync.py EMAIL PASSWORD hrv.json

To analyze this with PhysioNet's hrv toolkit, fetch two more git trees
into this tree:

    git clone https://github.com/kees/physionet-wfdb.git
    (cd physionet-wfdb && ./configure && make)

    git clone https://github.com/kees/physionet-hrv.git
    (cd physionet-hrv && git checkout fixes && make)

The RRs (in msec) can be fed into PhysioNet's hrv toolkit for analysis:

    for date in $(./rr-dump.py hrv.json | cut -d" " -f1); do
        ./rr-dump.py hrv.json $i > rr.dat
        echo -n $i" "
        (cd physionet-hrv; \
         LD_LIBRARY_PATH=../physionet-wfdb/build/lib \
         PATH=".:../physionet-wfdb/build/bin:$PATH" \
         ./get_hrv -mMR ../rr.dat 2>/dev/null) | grep rMSSD
    done

EliteHRV takes short reading, so be aware, quoting
http://physionet.org/tutorials/hrv-toolkit/

    Although the long term (24-hour) statistics of SDANN, SDNNIDX and
    ULF power can be calculated for shorter data lengths, they will
    become increasingly unreliable. For short-term data (less than 15
    minutes in length), only the time domain measures of AVNN, SDNN,
    rMSSD and pNN50 and the frequency domain measures of total power,
    VLF power, HF power and LF/HF ratio should be used.