System and method for automatic molecular diagnosis of ALS based on boosting classification

1. A method of diagnosing Amyotrophic lateral sclerosis comprising the steps of:

providing surface enhanced laser desorption/ionisation mass spectrometric (SELDI-MS) data of a plurality of proteins, said data obtained from a patient and comprising a plurality of peak values, wherein said SELDI-MS peak values are identified from a univariant analysis of SELDI-MS data obtained from a plurality of patients, including patients diagnosed with Amyotrophic lateral sclerosis and control patients; and

analysing said peak values with an alternating decision tree comprising a set of tests of said data peaks values and associated prediction values, wherein said data is predictive of Amyotropic lateral sclerosis if a sum of the prediction values of said tree is greater than 1.0.

2. The method of claim 1 , wherein said univariant analysis comprises performing a standard t-test and a bonferroni correction.

3. The method of claim 1 , wherein said alternating decision tree is trained by applying a boosting algorithm to the SELDI-MS peak values identified from said plurality of patients.

4. The method of claim 3 , wherein said boosting algorithm comprises the steps of:

providing a set of training examples (x 1 , y 1 ), . . . , (x m , y m ) where where x i ∈R d and y i ∈{−1,+1};

providing set of base conditions;

initialising a set of preconditions to TRUE;

initialising a set of weights for said training examples to a same initial value;

initialising said alternating decision tree to a single rule with a precondition and base condition both set to TRUE;

adding a new rule to said alternating decision tree, wherein said new rule comprises a base condition and a precondition that minimize a prediction function;

updating said prediction set with a conjunction of said base condition and a negation of said base condition; and

updating the set of weights based on prediction values of said new rule.

5. The method of claim 4 , wherein said prediction function Z t (c 1 , c 2 ), wherein c 1 represents a precondition and c 2 represents a base condition, is defined by

Z

t

⁡

(

c

1

,

c

2

)

=

2

⁢

(

W

+

⁡

(

c

1

⋀

c

2

)

⁢

W

-

⁡

(

c

1

⋀

c

2

)

+

W

+

⁡

(

c

1

⋀

⫬

c

2

)

⁢

W

-

⁡

(

c

1

⋀

⫬

c

2

)

)

+

W

⁡

(

⫬

c

2

)

,

wherein W + (c), W − (c) denote the total weight of training set examples that satisfy c.

6. The method of claim 4 , wherein the prediction values of said new rule are

1

2

⁢

ln

⁡

(

W

+

⁡

(

c

1

⋀

c

2

)

W

-

⁡

(

c

1

⋀

c

2

)

)

⁢

⁢

and

⁢

⁢

1

2

⁢

ln

⁡

(

W

+

⁡

(

c

1

⋀

⫬

c

2

)

W

1

⁡

(

c

1

⋀

⫬

c

2

)

)

,

wherein c 1 represents a precondition and c 2 represents a base condition, and wherein W + (c), W − (c) denote the total weight of training set examples that satisfy c.

7. The method of claim 6 , wherein updating the set of weights comprises multiplying each said weight by exp(−y i r(x i ) wherein r(x i ) represents a value associated with training example x i by the new rule r, and, and y i ∈{−1,+1}.

8. The method of claim 1 , wherein said proteins are obtained from the cerebrospinal fluid of a patient.

9. The method of claim 1 , wherein said plurality of proteins were absorbed on a weak cationic exchanger chip.

10. A method of diagnosing Amyotrophic lateral sclerosis comprising the steps of:

providing surface-enhanced laser desorption/ionisation mass spectrometric (SELDI-MS) data of a plurality of proteins, said data obtained from a plurality of patients, including patients diagnosed with Amyotrophic latcral sclerosis and control patients,

statistically analysing said data to identity significant peak values; and

training an alternating decision tree with a boosting algorithm using said significant peak values, wherein said alternating decision tree comprises a set of tests of said peaks values and associated prediction values.

11. The method of claim 10 , further comprising using said alternating decision tree to analyze SELDI-MS data obtained from a new patient.

12. The method of claim 11 , wherein an output of said alternating decision tree analysis of said new patient data is predictive of Amyotrophic lateral sclerosis if a sum of the prediction values of said tree is greater than 1.0.

13. A program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform the method steps for diagnosing Amyotrophic lateral sclerosis comprising the steps of:

providing surface-enhanced laser desorption/ionisation mass spectrometric (SELDI-MS) data of a plurality of proteins, said data obtained from a patient and comprising a plurality of peak values wherein said SELDI-MS peak values are identified from a univariant anal sis of SELDI-MS data obtained from a plurality of patients, including patients diagnosed with Amyotrophic lateral sclerosis and control patients; and

analysing said peak values with an alternating decision tree comprising a set of tests of said data peaks values and associated prediction values, wherein said data is predictive of Amyotrophic lateral sclerosis if a sum of the prediction values of said tree is greater than 1.0.

14. The computer readable program storage device of claim 13 , wherein said univariant analysis comprises performing a standard t-test and a bonferroni correction.

15. The computer readable program storage device of claim 13 , wherein said alternating decision tree is trained by applying a boosting algorithm to the SELDI-MS peak values identified from said plurality of patients.

16. The computer readable program storage device of claim 15 , wherein said boosting algorithm comprises the steps of:

providing a set of training examples (x 1 , y 1 ), . . . , (x m , y m ) where where x i ∈R d and y i ∈{−1,+1};

providing set of base conditions;

initialising a set of preconditions to TRUE;

initialising a set of weights for said training examples to a same initial value;

initialising said alternating decision tree to a single rule with a precondition and base condition both set to TRUE;

adding a new rule to said alternating decision tree, wherein said new rule comprises a base condition and a precondition that minimize a prediction function;

updating said prediction set with a conjunction of said base condition and a negation of said base condition; and

updating the set of weights based on prediction values of said new rule.

17. The computer readable program storage device of claim 16 , wherein said prediction function Z t (c 1 , c 2 ), wherein c 1 represents a precondition and c 2 represents a base condition, is defined by

Z

t

⁡

(

c

1

,

c

2

)

=

2

⁢

(

W

+

⁡

(

c

1

⋀

c

2

)

⁢

W

-

⁡

(

c

1

⋀

c

2

)

+

W

+

⁡

(

c

1

⋀

⫬

c

2

)

⁢

W

-

⁡

(

c

1

⋀

⫬

c

2

)

)

+

W

⁡

(

⫬

c

2

)

,

wherein W + (c), W − (c) denote the total weight of training set examples that satisfy c.

18. The computer readable program storage device of claim 16 , wherein the prediction values of said new rule are

1

2

⁢

ln

⁡

(

W

+

⁡

(

c

1

⋀

c

2

)

W

-

⁡

(

c

1

⋀

c

2

)

)

⁢

⁢

and

⁢

⁢

1

2

⁢

ln

⁡

(

W

+

⁡

(

c

1

⋀

⫬

c

2

)

W

1

⁡

(

c

1

⋀

⫬

c

2

)

)

,

wherein c 1 represents a precondition and c 2 represents a base condition, and wherein W + (c), W − (c) denote the total weight of training set examples that satisfy c.

19. The computer readable program storage device of claim 18 , wherein updating the set of weights comprises multiplying each said weight by y i r(x i )), wherein r(x i ) represents a value associated with training example x i by the new rule r, and, and y i ∈{−1,+1}.

20. The computer readable program storage device of claim 13 , wherein said proteins are obtained from the cerebrospinal fluid of a patient.

21. The computer readable program storage device of claim 13 , wherein said plurality of proteins were absorbed on a weak cationic exchanger chip.