コード例 #1
0
ファイル: real.py プロジェクト: yangzpag/conclave
def protocol_local(suffix: str, pid: int):
    pid_col_meds = "0"
    med_col_meds = "4"
    date_col_meds = "7"

    pid_col_diags = "8"
    diag_col_diags = "16"
    date_col_diags = "18"

    num_med_cols = 8
    num_diag_cols = 13

    left_medication_cols = [defCol(str(i), "INTEGER", pid) for i in range(num_med_cols)]
    medication = cc.create(suffix + "_medication", left_medication_cols, {pid})
    left_diagnosis_cols = [defCol(str(i + num_med_cols), "INTEGER", pid) for i in range(num_diag_cols)]
    diagnosis = cc.create(suffix + "_diagnosis", left_diagnosis_cols, {pid})

    shared_pids = cc.create("a_{}_shared_pids".format(suffix), [defCol(pid_col_meds, "INTEGER", pid)], {pid})

    # only keep relevant columns
    medication_proj = cc.project(medication, "medication_proj", [pid_col_meds, med_col_meds, date_col_meds])
    medication_mine = cc.filter_by(medication_proj, "medication_mine", pid_col_meds, shared_pids, use_not_in=True)

    diagnosis_proj = cc.project(diagnosis, "diagnosis_proj", [pid_col_diags, diag_col_diags, date_col_diags])
    diagnosis_mine = cc.filter_by(diagnosis_proj, "diagnosis_mine", pid_col_diags, shared_pids, use_not_in=True)

    joined = cc.join(medication_mine, diagnosis_mine, "joined", [pid_col_meds], [pid_col_diags])

    cases = cc.cc_filter(joined, "cases", date_col_diags, "<", other_col_name=date_col_meds)
    aspirin = cc.cc_filter(cases, "aspirin", med_col_meds, "==", scalar=1)
    heart_patients = cc.cc_filter(aspirin, "heart_patients", diag_col_diags, "==", scalar=1)

    cc.distinct_count(heart_patients, "actual_" + suffix, pid_col_meds)

    return {medication, diagnosis}
コード例 #2
0
ファイル: simple.py プロジェクト: yangzpag/conclave
def protocol():
    input_columns_left = [
        defCol("column_a", "INTEGER", [1]),
        defCol("column_b", "INTEGER", [1])
    ]
    left = cc.create("left", input_columns_left, {1})
    input_columns_right = [
        defCol("column_a", "INTEGER", [1]),
        defCol("column_b", "INTEGER", [1])
    ]
    right = cc.create("right", input_columns_right, {1})
    rel = cc.concat([left, right], "rel")
    filtered = cc.cc_filter(rel, "filtered", "column_b", "==", scalar=1)
    in_order = cc.sort_by(filtered, "in_order", "column_a")
    cc.distinct_count(in_order, "expected", "column_a")
    return {left, right}
コード例 #3
0
ファイル: real.py プロジェクト: yangzpag/conclave
def protocol(all_pids: list):
    pid_col_meds = "0"
    med_col_meds = "4"
    date_col_meds = "7"

    pid_col_diags = "8"
    diag_col_diags = "16"
    date_col_diags = "18"

    num_med_cols = 8
    num_diag_cols = 13

    left_medication_cols = [
        defCol(str(i), "INTEGER", 1) for i in range(num_med_cols)
    ]
    # public PID column
    left_medication_cols[0] = defCol(pid_col_meds, "INTEGER", all_pids)
    left_medication = cc.create("left_medication", left_medication_cols, {1})

    left_diagnosis_cols = [
        defCol(str(i + num_med_cols), "INTEGER", 1)
        for i in range(num_diag_cols)
    ]
    # public PID column
    left_diagnosis_cols[0] = defCol(pid_col_diags, "INTEGER", all_pids)
    left_diagnosis = cc.create("left_diagnosis", left_diagnosis_cols, {1})

    right_medication_cols = [
        defCol(str(i), "INTEGER", 2) for i in range(num_med_cols)
    ]
    # public PID column
    right_medication_cols[0] = defCol(pid_col_meds, "INTEGER", all_pids)
    right_medication = cc.create("right_medication", right_medication_cols,
                                 {2})

    right_diagnosis_cols = [
        defCol(str(i + num_med_cols), "INTEGER", 2)
        for i in range(num_diag_cols)
    ]
    # public PID column
    right_diagnosis_cols[0] = defCol(pid_col_diags, "INTEGER", all_pids)
    right_diagnosis = cc.create("right_diagnosis", right_diagnosis_cols, {2})

    medication = cc.concat([left_medication, right_medication], "medication")
    diagnosis = cc.concat([left_diagnosis, right_diagnosis], "diagnosis")

    # only keep relevant columns
    medication_proj = cc.project(medication, "medication_proj",
                                 [pid_col_meds, med_col_meds, date_col_meds])
    diagnosis_proj = cc.project(
        diagnosis, "diagnosis_proj",
        [pid_col_diags, diag_col_diags, date_col_diags])

    joined = cc.join(medication_proj, diagnosis_proj, "joined", [pid_col_meds],
                     [pid_col_diags])

    cases = cc.cc_filter(joined,
                         "cases",
                         date_col_diags,
                         "<",
                         other_col_name=date_col_meds)
    aspirin = cc.cc_filter(cases, "aspirin", med_col_meds, "==", scalar=1)
    heart_patients = cc.cc_filter(aspirin,
                                  "heart_patients",
                                  diag_col_diags,
                                  "==",
                                  scalar=1)

    cc.collect(cc.distinct_count(heart_patients, "actual", pid_col_meds), 1)

    return {left_medication, left_diagnosis, right_medication, right_diagnosis}
コード例 #4
0
ファイル: real.py プロジェクト: yangzpag/conclave
def protocol_mpc(all_pids: list):
    pid_col_meds = "0"
    med_col_meds = "4"
    date_col_meds = "7"

    pid_col_diags = "8"
    diag_col_diags = "16"
    date_col_diags = "18"

    num_med_cols = 8
    num_diag_cols = 13

    left_medication_cols = [defCol(str(i), "INTEGER", 1) for i in range(num_med_cols)]
    # public PID column
    left_medication_cols[0] = defCol(pid_col_meds, "INTEGER", all_pids)
    left_medication = cc.create("left_medication", left_medication_cols, {1})

    left_diagnosis_cols = [defCol(str(i + num_med_cols), "INTEGER", 1) for i in range(num_diag_cols)]
    # public PID column
    left_diagnosis_cols[0] = defCol(pid_col_diags, "INTEGER", all_pids)
    left_diagnosis = cc.create("left_diagnosis", left_diagnosis_cols, {1})

    right_medication_cols = [defCol(str(i), "INTEGER", 2) for i in range(num_med_cols)]
    # public PID column
    right_medication_cols[0] = defCol(pid_col_meds, "INTEGER", all_pids)
    right_medication = cc.create("right_medication", right_medication_cols, {2})

    right_diagnosis_cols = [defCol(str(i + num_med_cols), "INTEGER", 2) for i in range(num_diag_cols)]
    # public PID column
    right_diagnosis_cols[0] = defCol(pid_col_diags, "INTEGER", all_pids)
    right_diagnosis = cc.create("right_diagnosis", right_diagnosis_cols, {2})

    # Manual slicing
    left_keys = cc.union(left_medication, left_diagnosis, "left_pids", pid_col_meds, pid_col_diags)
    right_keys = cc.union(right_medication, right_diagnosis, "right_pids", pid_col_meds, pid_col_diags)

    left_shared_pids = cc._pub_intersect(left_keys, "a_left_shared_pids", pid_col_meds)
    cc._persist(left_shared_pids, "a_left_shared_pids")
    right_shared_pids = cc._pub_intersect(right_keys, "a_right_shared_pids", pid_col_meds, is_server=False)
    cc._persist(right_shared_pids, "a_right_shared_pids")

    left_medication_proj = cc.project(left_medication, "left_medication_proj",
                                      [pid_col_meds, med_col_meds, date_col_meds])
    left_medication_shared = cc.filter_by(left_medication_proj, "left_medication_shared", pid_col_meds,
                                          left_shared_pids)

    left_diagnosis_proj = cc.project(left_diagnosis, "left_diagnosis_proj",
                                     [pid_col_diags, diag_col_diags, date_col_diags])
    left_diagnosis_shared = cc.filter_by(left_diagnosis_proj, "left_diagnosis_shared", pid_col_diags, left_shared_pids)

    right_medication_proj = cc.project(right_medication, "right_medication_proj",
                                       [pid_col_meds, med_col_meds, date_col_meds])
    right_medication_shared = cc.filter_by(right_medication_proj, "right_medication_shared", pid_col_meds,
                                           right_shared_pids)

    right_diagnosis_proj = cc.project(right_diagnosis, "right_diagnosis_proj",
                                      [pid_col_diags, diag_col_diags, date_col_diags])
    right_diagnosis_shared = cc.filter_by(right_diagnosis_proj, "right_diagnosis_shared", pid_col_diags,
                                          right_shared_pids)

    # Slicing done
    medication_shared = cc.concat([left_medication_shared, right_medication_shared], "medication_shared")
    diagnosis_shared = cc.concat([left_diagnosis_shared, right_diagnosis_shared], "diagnosis_shared")

    joined = cc.join(medication_shared, diagnosis_shared, "joined", [pid_col_meds], [pid_col_diags])
    cases = cc.cc_filter(joined, "cases", date_col_diags, "<", other_col_name=date_col_meds)
    aspirin = cc.cc_filter(cases, "aspirin", med_col_meds, "==", scalar=1)
    heart_patients = cc.cc_filter(aspirin, "heart_patients", diag_col_diags, "==", scalar=1)

    cc.collect(cc.distinct_count(heart_patients, "actual_mpc", pid_col_meds), 1)

    return {
        left_medication,
        left_diagnosis,
        right_medication,
        right_diagnosis
    }